Republic of Iraq

Ministry of Higher Education and Scientific Research

University of Tikrit
College of Computer Science and Mathematics
Department of Mathematics
Post Graduate

Parameters Estimation and Properties of

New Odd Weibull-G Family with
Application
A Dissertation Submitted to

The Council of the College of Computer Science and Mathematics

Tikrit University

In Partial Fulfillment of the Requirements for the Degree of

Ph.D in Mathematics

By

Nooruldeen Ayad Noori Alsaab

Supervised by

Prof. Dr. Mundher Abdullah Khaleel

2025 A.D 1447 A.H.

 DEDICATION
To those whose light has illuminated the paths of my life,
To those who have been my support and the secret of my success,
To those who have devoted their lives to making mine more
beautiful,
I dedicate my work and achievement:
To Allah, first and foremost, Who guided and supported me
through His grace and generosity.
To the Prophet of Allah, Muhammad, peace and blessings be upon
him, the beacon of guidance and benevolence.
To my dear mother, the steadfast mountain, whose prayers became
the ladder to my ascent. Her gentle hand uplifted me, and her sincere
supplication was my shield.
To my compassionate father, who taught me that hardship fades
but knowledge endures, and that great achievements are built with
patience and kindness.
To my faithful wife, my companion on this journey—you are the
heart that embraced my dreams, and the serenity I turn to after every
hardship. You endured my absence, struggled silently by my side, and
your smile in my moments of exhaustion was the truest reward before any
accolade.
To my beloved daughters, Lina and Aya, the light of my eyes and
the flowers of my heart.
To my brothers and sister, who stood united with me in every step.
To everyone who extended a helping hand, through a kind word or
a sincere prayer.
If there is any good in this work, it is by the grace of Allah and
through their efforts.
If there is any shortcoming, it is from me and my own limitations.

Researcher …
 Acknowledgements

All praise is due to Allah, Lord of the Worlds, by whose

grace all righteous deeds are completed. Praise be to Him—
praise that equals His blessings and matches His bounty. May
peace and blessings be upon our master Muhammad, the teacher
of humanity and the beacon of knowledge, and upon his family
and all his companions.
As I place the final touches on this scholarly endeavor, I
find myself indebted with a gratitude that cannot be repaid
except through heartfelt thanks to my esteemed supervisor,
Professor Dr. Mundher Abdullah khaleel. You have been the
candle that lit my path, and the vessel that carried me through
the seas of research and knowledge. May Allah reward you with
the best recompense He grants to dedicated scholars.
I also extend my sincere appreciation to the respected
Dean, the esteemed Head of the Mathematics Department, and
all my honorable professors in the Department of Mathematics,
who were the flourishing garden from which I drew the essence
of knowledge, and the guiding light by which I found my way.
I entrust you to Allah, in whose care nothing is ever lost,
asking the Almighty to make this work sincerely for His noble
face and of benefit to His servants. Indeed, He is All-Hearing,
Responsive.

Researcher …
 Supervisor's Certification

I certify that the Dissertation entitled " Parameters Estimation and Properties of
New Odd Weibull-G Family with Application" has been prepared under my
supervision at the Department of Mathematics, College of Computer Science and
Mathematics, University of Tikrit, as a fulfilment of the requirement for the Degree
of Doctor of Philosophy in Mathematics.
Signature:
Name: Mundher Abdullah Khaleel
Scientific rank and certificate Prof. Dr.
Date: / /
Report of Linguistic Evaluation
I certify that the linguistic evaluation of this Dissertation was carried out by me and
it is linguistically acceptable and in expression.
Signature:
Name: Maha Bakir Mohammed
Scientific rank and certificate: Asst. Prof. Dr.
Date: / /
Report of Director of Postgraduate Studies Committee
According to the recommendations presented by the supervisor and the linguistic
evaluator of this Dissertation, I recommend that this Dissertation be forwarded for
discussion.
Signature:
Name: Mundher Abdullah Khaleel
Scientific rank and certificate: Prof. Dr.
Date: / /
Report of the Head of the Department of Mathematics
According to the Recommendations presented by the supervisor and the linguistic
evaluator of this Dissertation and the director of postgraduate studies committee, I
recommend that this Dissertation be forwarded for discussion.
Signature:
Name: Ahmed Maher Salih
Scientific rank and certificate: Teacher Dr.
Date: / /
 EXAMINATION COMMITTEE CERTIFICATION

We certify that we have read the Dissertation entitled "Parameters

Estimation and Properties of New Odd Weibull-G Family with Application",
and examining committee examined the student (Nooruldeen Ayad Noori) in its
contents and what is related with it in 11/08/2025, and that in our opinion it is
adequate with (Excellent) standing as a Dissertation for the degree Doctor of
Philosophy of Science in Mathematics.
Signature: Signature:
Name: Mizal Hamad Alobaidi Name: Sudad Khalil Abraheem
Scientific rank and certificate: Prof. Dr. Scientific rank and certificate: Prof. Dr.
Date: / / 2025 Date: / / 2025
(Chairman) (Member)
Signature: Signature:
Name: Azher Abas Mohammad Name: Qasim Nasir Husain
Scientific rank and certificate: Asst. Prof. Dr. Scientific rank and certificate: Asst. Prof. Dr.
Date: / / 2025 Date: / / 2025
(Member) (Member)
Signature: Signature:
Name: Ghadeer Jasim Mohammed Name: Mundher Abdullah Khaleel
Scientific rank and certificate: Asst. Prof. Dr. Scientific rank and certificate: Prof. Dr.
Date: / / 2025 Date: / / 2025
(Member) (Member and Supervisor)

Authentication of Dean of the College of Computer Science and Mathematics

The Council of the college of Computer Science and Mathematics met in its
session ________ held on / / 2025 and decided to award him a degree Doctor of
Philosophy of Science in Mathematics.

Signature:
Name: Asst. Prof. Dr. Mahmood Maher Salih
Computer Science and Mathematics College Dean
 ABSTRACT

Abstract
This study aims to introduce a new probability distribution family
known as the " New Odd Weibull-G Family." It is an extension of the
classical Weibull distribution, combining it with hybrid generation
techniques to produce more flexible models that can represent real-world
data, both normal and neutrosophic. This study addresses the
shortcomings of traditional distributions in representing asymmetric data
or data containing a high degree of uncertainty, particularly in
applications of survival analysis, medical data, and financial data.
The study involved the development of two distribution models
within the proposed family: one for normal data, and the second for
neutrosophic data. The basic mathematical properties of the distributions
were derived, such as the probability density function, cumulative
distribution function, reliability function, and moment functions. Model
parameters are estimated using three different methods, including
maximum likelihood and least squares, and the efficiency of each method
was evaluated through Monte Carlo simulation.
The models are applied to real data using the R software, and their
performance is compared with other distributions according to
information criteria such as Akaike information criteria (AIC), and
Bayesian information criteria (BIC). The results demonstrated that the
new distribution outperforms data representation, especially when
uncertainty exists or data are heavily tailed. The study also included tools
to check the degree of uncertainty in the data before modeling to ensure
the reliability of the analysis.

I
 LIST OF ABBREVIATIONS

List of Abbreviations
Abbreviation Meaning
T Truth
F False
I Indeterminacy
AIC Akaike Information criterion
AICC Correct Akaike Information criterion
BIC Bayesian Information criterion
HQIC Hannan and Quinn Information criterion
𝑲𝑺 Kolmogorov-Smirnov Statistics
𝑰(∗) indicator function
A Anderson-Darling Statistics
W Cramer-Von Mises Statistics
p-value Probability Value Statistics
NOWG New Odd Weibull-G family
𝑸(𝒖) quantile function
𝒒 𝒒
𝑾 (𝒑; 𝒃) Lombart function for (𝒑; 𝒃)

𝑾−𝟏 (𝜷𝒆−𝜷 ) lower Lombart function for 𝜷𝒆−𝜷

𝛍𝒏 𝒏𝒕𝒉 moment of any distribution
𝒎𝒈𝒇 moment generating function
𝑰𝑹 (𝒄)𝑯𝑾𝑮 Rényi entropy for the NOWG Family
NOWIBX New Odd Weibull Inverse Burr Type X
Expectation for 𝐗~𝐪, and 𝒒 is standard exponential
𝑬𝒒 (∗)
distribution

..

III
 LIST OF ABBREVIATIONS
MLE Maximum Likelihood Estimation
OLSE Ordinary Least Squares Estimation
WLSE Weighted Least Squares Estimation
RMSEs Root mean squared errors
𝑻𝜶 True membership function
𝑭𝜶 False membership function
𝑰𝜶 Undefined membership function
𝑿𝑵 Neutrosophic random variable
NeCDF Neutrosophic Cumulative Function
NePDF Neutrosophic density function
NOWIR New Odd Weibull Inverse Rayleigh distribution
Neutrosophic New Odd Weibull Inverse Rayleigh
NeNOWIR
distribution
𝛍̀ 𝖒 𝑵𝒆 Neutrosophic 𝒌𝒕𝒉 moments
𝑺𝑲𝑵𝒆 Neutrosophic skewness
𝑲𝑼𝑵𝒆 Neutrosophic kurtosis
𝝈𝟐𝑵𝒆 Neutrosophic variance

IV
 LIST OF FIGURES

List of Figures
No. The Title Page No.
3.1 CDF function of NOWIBX distribution for different
34
parameter values
3.2 pdf function of NOWIBX distribution for different
35
parameter values
3.3 Survival function of NOWIBX distribution for
36
different parameter values
3.4 Hazard function of NOWIBX distribution for different
37
parameter values
3.5 Fitted densities for Data I 61
3.6 Fitted empirical CDF for Data I 61
3.7 Fitted densities for Data II 64
3.8 Fitted empirical CDF for Data II 64
4.1 NeCDF for NeNOWIR distribution 68
4.2 Nepdf for NeNOWIR distribution 69
4.3 survival for NeNOWIR distribution 70
4.4 Hazard for NeNOWIR distribution 71
4.5 (a) 3D plot of Variance, (b) 3D plot of Skewness, (c) 3D
78
plot of Kurtosis
4.6 plot the intervals for data-I used 91
4.7 Histogram of Neutrosophic Components 91
4.8 Neutrosophic parts and sum of Neutrosophic
92
Components
4.9 Fitting pdfs NeNOWIR with histogram data-I 95
4.10 Empirical Fitted CDFs NeNOWIR with histogram
95
data-I
4.11 plot the intervals for data used-II 98
4.12 Histogram of Neutrosophic Components 99

V
 LIST OF FIGURES

4.13 Neutrosophic parts and sum of Neutrosophic Components 99

4.14 Fitting pdfs NeNOWIR with histogram data-II 101
4.15 Empirical Fitted CDFs NeNOWIR with histogram data-II 102

VI
 LIST OF TABLES

List of Tables
No. The Title Page No.
1.1 Comparison between the three tests (KS, W, A) 12
2.1 Some individuals in the T-X method 21
The quantiles for some parameter values of NOWIBX
3.1 41
distribution
Numerical value of 𝝁𝟏 , 𝝁𝟐 , 𝝁𝟑 , 𝝁𝟒 , 𝝈𝟐 , 𝑺𝑲, and 𝑲𝑼 of the
3.2 44
NOWIBX distribution
3.3 Monte Carlo simulations-1 conducted for the NOWIBX 52
3.4 Monte Carlo simulations-2 conducted for the NOWIBX 54
3.5 CDF functions for comparative distributions 58
Information criteria values for comparative distributions
3.6 59
for data I
Evaluate goodness-of-fit statistics for comparative
3.7 59
distributions for data I
parameter estimators by MLE for comparative
3.8 60
distributions for data I
Information criteria values for comparative distributions
3.9 62
for data II
Evaluate goodness-of-fit statistics for comparative
3.10 63
distributions for data II
Parameter estimators by MLE for comparative
3.11 63
distributions for data II
Quantile function values for different intervals for
4.1 74
𝒂𝑵𝒆 , 𝒃𝑵𝒆 and 𝒄𝑵𝒆
Numerical value of 𝝁̀ 𝟏 𝑵𝒆 , 𝝁̀ 𝟐 𝑵𝒆 , 𝝁̀ 𝟑 𝑵𝒆 , 𝝁̀ 𝟒 𝑵𝒆 , 𝝈𝟐𝑵𝒆 , 𝑺𝑲𝑵𝒆 , and
4.2 77
𝑲𝑼𝑵𝒆 of the NeNOWIR distribution
Monte Carlo simulations for statistical method conducted
4.3 84
for NeNOWIR
Monte Carlo simulations for numerical optimization
4.4 86
conducted for NeNOWIR
4.5 Comparative distributions 89
4.6 Data-1 used, Truth, False, and indeterminacy values 90
4.7 Descriptive statistics for the data-I 92
4.8 Estimates of models for data-I 93
4.9 Evaluate statistical metrics for the data-I 93
4.10 parameter estimators by MLE for the data-I 94
4.11 Data used-II, Truth, False, and indeterminacy values 96
4.12 Descriptive statistics for the data-II 99
4.13 Estimates of models for data-II 99
4.14 Evaluate statistical metrics for the data-II 99
4.15 parameter estimators by MLE for the data-II 100

VII
 LIST OF THE CONTENTS

List of the Contents

No. Subjects Page No.
--- Abstract I
--- List of abbreviations III
--- List of Figures V
--- List of Tables VII
--- List of the Contents VIII
--- Introduction 1-8
Chapter One: Some Basic Concepts and
1 9-18
related Definitions
1.1 Some Basic Definitions 9
1.2 Neutrosophic Logic 12
2 Chapter Two: A New Odd Weibull-G family
19-32
with some its properties
2.1 Introduction 19
2.2 The New Odd Weibull-G (NOWG) family 19
2.3 Some Properties of NOWG family 22
3 Chapter Three: A New Odd Weibull Inverse
33-65
Burr Type X, properties, with application
3.1 Introduction 33
New Odd Weibull Inverse Burr-X (NOWIBX)
3.2 33
distribution
Mathematical Properties of NOWIBX
3.3 38
distribution
3.4 Estimation 47
3.5 Simulation Study 51
3.6 Application 57
Chapter four: Neutrosophic New Odd Weibull
4 66-102
inverse Rayleigh Distribution
4.1 Introduction 66
4.2 NOWIR distribution 67
Neutrosophic New Odd Weibull Inverse
4.3 67
Rayleigh distribution
Statistical Properties for NeNOWIR
4.4 70
distribution
4.5 Estimation 82
4.6 Simulation 83
4.7 Application 88
5 Chapter five: Conclusions and 103-105

VIII
 LIST OF THE CONTENTS

Recommendations
5.1 Conclusions 103
5.2 Recommendations 104
--- References 106-112
--- ‫المستخلص‬ ---

IX
 INTRODUCTION

Over the past few decades, statistical modeling using classical

probability distributions has been in the field of statistical distributions
and generating families of continuous distributions witnessed significant
development in the use of classical probability distributions to model data
in various fields, such as engineering, environmental and medical
sciences, biological and demographic studies, economic and insurance.
However, there has been an urgent need to expand these distributions to
meet the requirements of practical applications, especially in life
expectancy analysis and insurance.
The primary challenge confronting researchers lies in the limitations
of traditional probability distributions, such as the normal and exponential
distributions, in adequately modeling diverse types of data., to accurately
represent non-normal or asymmetric data. For example, data exhibiting
long-tailed or multi-peaked distributions require more flexible models.
Time-restricted data or data concentrated at one tail are examples of
situations that require the development of new distribution families.
To overcome these challenges, several techniques have emerged to
generate new families of probability distributions. Among these methods
is the Beta-G method, introduced by Eugene et al. in 2002 [1], which is
based on beta distribution function and allows greater flexibility through
additional shape parameters. The T-X method was developed by
Alzaatreh et al. in 2014 [2], which led to the generation of new families.
The new distributions resulting from these methods provide
significant improvements in modeling complex data, allowing for a
deeper understanding of statistical phenomena and more accurate analysis
of unconventional data. This development enhances the efficiency of
statistical tools and enables researchers to handle diverse data in multiple
fields, including the social sciences, medicine, and economics.

1
 INTRODUCTION

In addition, modern methods include the use of mixed and hybrid

distributions, which combine several conventional or extended
distributions to provide more accurate and flexible models. For example,
the lognormal distribution can be combined with the gamma distribution
to interpret data containing high variance or asymmetry. Furthermore, the
use of machine learning and artificial intelligence techniques to improve
parameter estimation and determine the optimal shape of distribution has
become a recent trend in this field.
Advances in cloud computing and big data analysis have also opened
up opportunities to apply these distribution families to large, complex,
real-time datasets. For example, in biomedicine, extended distributions
are used to analyze large-scale genomic data and understand rare genetic
patterns. In economics, these distributions help model financial markets
and predict price dynamics and volatility.
This progress has led to emergence of hundreds of new statistical
families in the current literature, which are difficult to enumerate due to
the diversity of methods used and the practical applications they can
leverage. This field is now one of the most active and evolving areas of
research, as researchers continue to refine models and develop various
scientific and industrial sectors.
In the field of neutrosophic logic and neutrosophic statistical
distribution, neutrosophic logic is an advanced extension of fuzzy logic,
introduced by Florin Smarandache 1999 [3]. It aims to deal with
uncertain, incomplete, and inconsistent systems. Unlike traditional logic,
which relies on binary values (True or False), and fuzzy logic, which
allows degrees of membership ranging from 0 to 1, neutrosophic logic
introduces three main components: Truth (T), non-Truth (False) (F), and
indeterminacy (I). These components allow for a more comprehensive
and flexible representation of uncertainty and contradiction.

2
 INTRODUCTION

In the statistical context, neutrosophic probability distributions

emerged as an extension of classical distributions to deal with data
containing uncertainty, contradiction, and high variance. Neutrosophic
distribution relies on defining probability in a way that takes into account
not only degree of belonging but also a degree of uncertainty and lack of
knowledge. For example, when analyzing medical data that contain
uncertain or contradictory results, neutrosophic distribution can provide
more accurate models and a realistic interpretation of the data situation.
Advances in this field include improving parameter estimation
techniques and developing computational methods based on artificial
neural networks and artificial intelligence to effectively determine
neutrosophic values. This approach is used in fields such as reliability
analysis, financial forecasting, and medical diagnosis, where data are
incomplete or affected by uncertain factors. Neutrosophic probability
distribution represents an important step toward enhancing the ability of
statistical models to handle the complexity of data in real-world
environments, opening up vast horizons for analysis of big data complex
dynamical systems [4].
This study proposes a hybrid of statistical distributions called New
Odd Weibull-G family that deals with both clinical and neutrosophic
data. Several mathematical properties of the proposed family are
presented. Two distributions are then presented within this family: one
for normal data and the other for neutrosophic data, to demonstrate the
flexibility of the proposed family. The most important statistical
properties of the proposed distribution are then found and studied. The
parameters are then estimated, and the efficiency of the parameter
estimation methods is demonstrated using Monte Carlo simulation.
Finally, to demonstrate the suitability of the proposed distributions for
modeling survival data, real data were used and analyzed using R

3
 INTRODUCTION

software. The distribution was then compared with some other

distribution to assess the best-fitting model.
Researchers Alzaatreh et al. (2013) introduced the T-X method to
generate new families of continuous distributions. This method relies on
applying a transformation to a random variable from underlying
distribution and expanding it using an additional function. The goal is to
improve the resilience of traditional distribution to handle asymmetric
and heavy-tailed data, with applications in reliability and risk analysis [2].
Many researchers have adopted this method as a methodology for their
work, for example:
In 2014, G. M. Cordeiro et al. Developed the Exponentiated Half-
Logistic family, which enhances the resilience of the half-logistic
distribution by raising the cumulative exponential function of the logistic
function to a certain power. This family was applied in reliability and
biological data analysis [5].

In 2015, G. M. Cordeiro et al. introduced the Generalized Weibull

family, which allows for the extension and improvement of the original
distribution through additional parameters. This family aims to improve
the fit of data with variable patterns and heavy tails [6].

M. H. Tahir et al. (2016) developed the New Weibull-G family, which

expands the Weibull distribution with additional parameters to control
shape and dispersion, making it suitable for analyzing reliability and
lifespan data [7].

J. Farrukh et al. (2017) developed the Odd Burr-III family, which is

based on Burr-III distribution and is used for the analysis of reliability
and environmental data [8].

4
 INTRODUCTION

M. Alizadeh et al. (2017) introduced the Gompertz-G family, which

expands the Gompertz distribution with additional parameters to improve
data fit in areas such as reliability and ecological analysis [9].

A. D. C. Nascimento et al. (2018) introduced the Odd Nadarajah-

Haghighi family, which adds flexibility to the Nadarajah-Haghighi
distribution for analysis of environmental data [10].

F. A. Bhatti et al. (2019) developed the Burr-III-Marshall Oklin

family, which combines Burr-III distribution with Marshall Oklin
distribution to improve the analysis of heavy-tailed data [11].

M. S. Eliwa et al. (2020) introduced the discrete Gompertz-G family

for analyzing over or under-distribution data. They focused on
mathematical properties, parameter estimation, and applications in
various fields such as economics and health sciences [12].

A. S. Hassan et al. (2021) introduced a New Topp-Leone-G family

based on Topp-Leone distribution to expand the scope of statistical
applications in the physical sciences [13].

M. Alizadeh et al. (2022) developed the Odd Log-Logistic Lindley-G

family, which combines the Lindley distribution with an asymmetric
logistic transformation. They used Bayesian and non-Bayesian estimation
to provide more accurate solutions for data analysis [14].

Z. Shah et al. (2023) introduced the generalized Logarithmic-X family

for biomedical data analysis, aiming to improve chronic disease
prediction [15].

G. A. Mahdi et al. (2024) developed the Hybrid Odd Exponential-Φ

family, which combines the exponential distribution with the Φ
transformation to provide greater flexibility in data analysis. This family

5
 INTRODUCTION

is used in fields such as physics and engineering to improve the

prediction accuracy of statistical models [16].

The study gap

gaps remain in developing more flexible distributions capable of
handling both normal and neutrosophic data sets. Traditional statistical
models often fail to characterize data containing uncertainty, requiring
more sophisticated models. There is no family of distributions has been
developed that combines the Weibull distribution and the hybrid
methodology to create new distributions that can handle both normal and
neutrosophic data simultaneously. Furthermore, most previous studies
focused on estimating distribution parameters using traditional methods
such as maximum distribution parameters using traditional methods such
as maximum likelihood, without exploring multiple estimation methods
to more deeply measure the efficiency of the models. Furthermore, the
extent of neutrosophicity of the data has not been systemically verified
before applying statistical models, which can lead to inaccurate
conclusions.
Accordingly, this dissertation seeks to fill this gap by:
1. Introducing a new family of distributions that combines the
Weibull distribution and the hybrid methodology, enhancing the
flexibility of statistical models.
2. Developing two sub-models within the new family: one for normal
data, and the other for neutrosophic data.
3. Comparing three different methods for estimating the parameters
(MLE, OLSE, WLSE) of the two distributions through Monte
Carlo simulations to evaluate their effectiveness.

6
 INTRODUCTION

4. Verifying the neutrosophicity of the data before applying the

model, ensuring the accuracy of the analysis and the reliability of
the results.
The new distribution was applied to real data to examine its
consistency with the actual data to provide practical insights into its
potential for characterizing uncertain data.

The significance of this study

Lies in presenting an integrated statistical framework that combines
the flexibility to describe normal data with the ability to analyze uncertain
data using a neutrosophic approach. This research contributes to
expanding the scope of the use of probability distribution in statistical
analysis, making it useful in multiple fields such as economics, medical
sciences, and risk analysis, where uncertainty plays a crucial role. The
predicted results also enable improved estimation and simulation
methods, enhancing the accuracy of analysis and decision-making based
on more realistic data.

Study Aim
This study aims to introduce a new family of distribution and study
some of their statistical properties. This family relies on combining the
Weibull distribution with a hybrid methodology to construct families of
continuous distributions based on finding a hybrid integral bound that
combines two T-X bounds. This proposed family, known as the NOWG,
also derives some of the proposed family’s statistical properties. This
proposed family introduces two new statistical distributions: one is
designed to handle normal data and, other to analyze Neutrosophic data,
with parameters estimated using five different methods. Simulations are
conducted to study the efficiency of these methods, along with practical
applications on real data. The other distribution is designed to handle

7
 INTRODUCTION

neutrosophic data, extending its scope by setting upper and lower bounds
for its parameters and random variables. Its statistical properties are
determined and applied to two types of neutrosophic data.

Dissertation structure
The Dissertation consists of five chapters as follows:
1. Chapter One: Introduces some basic definitions and concepts
about compatibility criteria and some statistical measures.
2. Chapter Two: Introduces the new family, which relies on
combining the Weibull distribution with a hybrid methodology,
and derives some basic properties of the proposed family
3. Chapter Three: Introduces a new distribution based on the
proposed family, using Inverse Burr Type X as the underlying
model with derives some statistical properties of this distribution.
Estimates the distribution parameters using three different methods
and Conducts a simulation to study the efficiency of the five
estimation methods and applies the proposed distribution to a set of
real data to test its consistency with normal data and evaluating its
performance experimentally.
4. Chapter Four: Introduces a new distribution based on the
proposed family, using Inverse Rayleigh and Neutrosophic logic.
Estimates the distribution parameters using some different
methods, similar to chapter three, and applies the proposed
distribution to real neutrosophic data.
5. Chapter Five: It includes the most important conclusions reached
by the Dissertation, along with some recommendations for future
studies.

8
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

Chapter One
Some Basic Concepts and related Definitions

1.1 Some Basic Definitions

In this part, we will discuss some of the most important informational
criteria for comparing distribution, which will be addressed during our
study of the practical aspect. The statistical model is chosen to be
appropriate or suitable for the data set used and compared so that the
distribution is chosen based on the data and results of these criteria. A
lower value for criterion indicates that the distribution has the best fit for
the data used.

1.2.1 Akaike Information criterion (AIC)

This criterion was first introduced by Akaike in 1987 as an estimate of
the expected relative divergence. It's one of the most common criteria and
is abbreviated as AIC. It is defined according to the following equation
[17]:
𝑁

𝐴𝐼𝐶 = 2 (𝑘 − ∑ log 𝑓𝑚 (𝑥𝑛 ⁄𝜗̂𝑚 )) (1.1)

𝑛=1

where 𝑓𝑚 is the pdf function, 𝜗̂𝑚 represents the maximum likelihood

estimator, 𝑘 is the number of parameters, and 𝑁 is the number of
observations.

1.2.2 Corrected Akaike Information criterion (AICC)

This criterion was first introduced by Hurvih and Tsai in 1989.
Because the AIC can appear quite biased, leading to severe overfitting,
they presented a corrected version of AIC that is more neutral and
therefore a better model compared with the AIC. It is abbreviated as
CAIC. Its defined according to the following equation [18]:

9
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

4𝑘 2 + 2𝑘
𝐴𝐼𝐶𝐶 = 𝐴𝐼𝐶 + (1.2)
𝑛−𝑘−1
Where 𝑛 is the sample size.

1.2.3 Bayesian Information criterion

In 1978, the researcher Schwasz proposed a new Information criterion,
symbolized by the symbol BIC. This criterion addresses the problems of
the lack of consistency that AIC suffers from. The formula for this
criterion [19] is:
𝑁

𝐵𝐼𝐶 = 𝑘 log 𝑛 − 2 ∑ log 𝑓𝑚 (𝑥𝑛 ⁄𝜗̂𝑚 ) (1.3)

𝑛=1

1.2.4 Hannan and Quinn Information criterion

This criterion was introduced by Hannan and Quinn's in 1979,
symbolized by the symbol JQIC. It is an alternative to AIC and BIC. The
formula for this criterion [19] is:
𝑁

𝐻𝑄𝐼𝐶 = 2𝑘 log[log 𝑛] − 2 ∑ log 𝑓𝑚 (𝑥𝑛 ⁄𝜗̂𝑚 ) (1.4)

𝑛=1

1.2.5 Kolmogorov-Smirnov Statistics (𝑲𝑺)

Kolmogorov-Smirnov Statistics used to test how closely a data sample
fits a given distribution. This statistic measures the maximum difference
between sample's empirical distribution 𝐹𝑛 (𝑥 ) and hypothesized
theoretical CDF 𝐹 (𝑥 ). It can be calculated using this formula [20]:
𝐾𝑆 = sup|𝐹𝑛 (𝑥 ) − 𝐹 (𝑥 )| (1.5)
𝑥

where 𝐹𝑛 (𝑥 ) is given by:

𝑁
1
𝐹𝑛 (𝑥 ) = ∑ 𝐼 (𝑋𝑖 ≤ 𝑥 ) (1.6)
𝑁
𝑖=1

where 𝐼 (∗) is the indicator function which takes value 1 if 𝑋𝑖 ≤ 𝑥, and

zero otherwise.

10
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

The smaller 𝐾𝑆 when the difference between 𝐹𝑛 (𝑥 ) and 𝐹 (𝑥 ) is

low. If the value of 𝐾𝑆 exceeds a certain threshold at a significance level
of 𝛼, the hypoDissertation that the data follow the assumed distribution is
rejected.

1.2.6 Anderson-Darling Statistics (A)

Anderson-Darling Statistics is an improvement over the 𝐾𝑆 test, given
greater weight to distribution difference at the tails (extreme values). The
mathematical equation for the statistic is expressed as follows [20]:
𝑛
2𝑖 − 1
𝐴 = −𝑛 − ∑ [log 𝐹 (𝑋𝑖 ) + log(1 − 𝐹 (𝑋𝑛+1−𝑖 ))] (1.7)
𝑛
𝑖=1

where 𝑋𝑖 is the order value for sample, and 𝐹 (𝑋𝑖 ) is theoretical CDF of a
sample.
If 𝐴 is the value of high, it means that the sample does not follow the
assumed distribution. The calculated value is compared to critical values
to obtain the test result.

1.2.7 Cramer-Von Mises Statistics (W)

Cramer-Von Mises Statistics used to measure how close the empirical
distribution of a sample is to a hypothesized distribution, but it distributes
the weights equally across all values (Unlike A test). It is calculated as
follows [20]:
𝑛 2
2𝑖 − 1 1
𝑊 = ∑ (𝐹(𝑋𝑖 ) − ) + (1.8)
2𝑛 12𝑛
𝑖=1

If the value of W is large, it indicates a poor fit between the data and
assumed distribution.

1.2.8 Probability Value Statistics (p-value)

Probability Value Statistics is the probability associated with a test
statistic. It represents the probability of obtaining an equal or more

11
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

extreme statistical value under the null hypothesis 𝐻0 . The basic equation
is [21]:
𝑝 = 𝑃 (𝑇 ≥ 𝑇𝑜𝑏𝑠 |𝐻0 ) (1.9)
where 𝑇 is the statistic used in the test (such as 𝐴, 𝐾𝑆, 𝑊), 𝑇𝑜𝑏𝑠 is the
calculated value from the sample, and 𝐻0 is the null hypothesis (that the
sample follows the assumed distribution).
If the p-value is less than the chosen significance level 𝛼 (usually
0.05), the 𝐻0 is rejected, meaning that the data do not follow the assumed
distribution.
Table (1.1): Comparison between the three tests (𝐾𝑆, W, A)
Test How to measure Main advantage weakness
Maximum difference between
empirical distribution 𝐹𝑛 (𝑥 ) and Simple and easy to Less accurate at
KS
hypothesized theoretical CDF interpret the tail
𝐹 (𝑥 )
Measures cumulative Less common
Addresses some of
W differences equally across the in practical
the problems of KS
range applications
Sensitive to More complex
A Gives more weight to the tails
outliers than KS

1.2 Neutrosophic Logic

It was introduced by Florentin Smarandache in 1999 [3], and it is an
extension of classical and fuzzy logic that provides a more nuanced
approach to dealing with uncertainty and imprecision in information. It's
grounded in concepts of Neutrosophic, which are weeks to address
limitations of traditional logic by introducing a components framework
composed of three independent components:
1. Truth (T): The degree to which a statement is true.
2. Falsity (F): The degree to which a statement is false.

12
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

3. Indeterminacy (I): The degree to which statement is uncertainty or

ambiguity associated with a statement, recognizing that some
information could be half true. Partially false, or neither.
Smaradache 2003 conducted an analytical investigation on
neutrosophic logic, a novel philosophical framework take seeks to
enhance human comprehension of fundamental cognitive truths [3], as
well as 2004 conducted a study on neutrosophic sets, as an extension for
intuitive fuzzy sets [22], while Georgiev 2005 proposed a new model
based on neutrosophic sets [23]. Patro and Smaradache 2016 introduced
an approach to incorporate neutrosophic distributions into statistical
theory. Here are of examples for neutrosophic statistical distributions: a
study in 2021 was conducted on statistical distributions based on this
logic [24]. Another study deals with theory of random variables utilizing
neutrosophic concepts [25]. These studies were supported by other
studies focused on presenting neutrosophic statistical distributions, for
example: neutrosophic-Lindley (NLD) [26], neutrosophic-Exponential-
inverse-Rayleigh (NEIRD), neutrosophic-Lindley (NLD) [27],
neutrosophic-inverse-power- Rayleigh (NIPLD) [28], Neutrosophic
Topp-Leone Extended Exponential (NTLEED) [29], and Neutrosophic
Beta-Lindley (NBL) [30].

1.2.1 Neutrosophic Inferential Statistics

Neutrosophic Inferential Statistics consists of methods that permit the
generalization from a neutrosophic sampling to a population from which
it was selected the sample [24].

1.2.2 Neutrosophic Data

Neutrosophic Data is the data that contains some indeterminacy.
Similarly to the classical statistics it can be classified as [24]:

13
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

1. Discrete neutrosophic data, if the values are isolated points; for

example: 6 + 𝑖1, where 𝑖1 ∈ [0,1].
2. Continuous neutrosophic data, if the values form one or more
intervals, for example: [0,0.8] or [0.1,1.0] (i.e. not sure which
one).

Another classification:
1. Quantitative (numerical) neutrosophic data; for example: a number
in the interval [2, 5] (we do not know exactly), 47, 52, 67 or 69
(we do not know exactly);
2. Qualitative (categorical) neutrosophic data; for example: blue or
red (we don’t know exactly), white, black or green or yellow (not
knowing exactly).
Also, we may have: -univariate neutrosophic data, i.e. neutrosophic
data that consists of observations on a neutrosophic single attribute; - and
multivariable neutrosophic data, i.e. neutrosophic data that consists of
observations on two or more attributes.
As a particular cases we mention the bivariate neutrosophic data, and
trivariate neutrosophic data.

1.2.3 Neutrosophic Statistical Number

A Neutrosophic Statistical Number 𝑁 has the form [24]:
𝑁 = 𝑑+𝑖
where 𝑑 is the determinate (sure) part of 𝑁, and 𝑖 is the indeterminate
(unsure) part of N.
For example, 𝑎=5+𝑖, where 𝑖 ∈ [0,0.4],is equivalent to 𝑎 ∈ [5,5.4],
so for sure 𝑎 ≥ 5 (meaning that the determinate part of a is 5), while the
indeterminate part 𝑖 ∈ [0,0.4] means the possibility for number 𝑎 to be a
little bigger than 5.

14
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

1.2.4 Neutrosophic Frequency Distribution

A Neutrosophic Frequency Distribution is a table displaying the
categories, frequencies, and relative frequencies with some
indeterminacies. Most often, indeterminacies occur due to imprecise,
incomplete or unknown data related to frequency. As a consequence,
relative frequency becomes imprecise, incomplete, or unknown too [24].

1.2.5 Neutrosophic Sample

A Neutrosophic Sample is a chosen subset of a population, subset that
contains some indeterminacy: either with respect to several of its
individuals (that might not belong to the population we study, or they
might only partially belong to it), or with respect to the subset as a whole.
While the classical samples provide accurate information, the
neutrosophic samples provide vague or incomplete information.
By language abuse one can say that any sample is a neutrosophic
sample, since one may consider their determinacy equals to zero [24].

1.2.6 Neutrosophic Survey

A Neutrosophic Survey Results are survey results that contain some
indeterminacy [24].

1.2.7 Neutrosophic Population

A Neutrosophic Population is a population not well determined at the
level of membership (i.e. not sure if some individuals belong or do not
belong to the population) [24].

1.2.8 Neutrosophic random sample

A neutrosophic random sample of size n from a classical or
neutrosophic population is a sample of n individuals such that at least one
of them has some indeterminacy [24].

15
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

1.2.9 Neutrosophic Real or Complex Polynomial.

A polynomial whose coefficients (at least one of them containing 𝐼)
are neutrosophic numbers is called Neutrosophic Polynomials.
Similarly we may have Neutrosophic Real Polynomials if its
coefficients are neutrosophic real numbers, and Neutrosophic Complex
Polynomials if its coefficients are neutrosophic complex numbers [24].

1.2.10 Neutrosophic Random Numbers

A Neutrosophic Random Numbers is a sequence of numbers and
indeterminacies occurring at random with equal probability. The
occurrence of a number or indeterminacy is not a guide to the numbers or
indeterminacies that follow it, nor is it predicted from the numbers or
indeterminacies that precede it [24].

1.2.11 Neutrosophic Weighted Random Numbers

A Neutrosophic Weighted Random Numbers, where each number 𝑥𝑗
has a different chance 𝑝𝑗 to occur, and each indeterminacy 𝐼𝑗 has a
different chance 𝑟𝑗 to occur. There are also cases when the numbers have
to be in a given set; for example, each number should have 𝑘 digits [24].

1.2.12 Component Neutrosophic Logic

Neutrosophic logic is an extension of fuzzy logic and axiomatic fuzzy
logic, where the logical variable 𝑋 is described in neutrosophic logic as a
triple [25]:
𝑋 = (𝑇, 𝐹, 𝐼 ) (1.10)
where 𝑇, 𝐹, and 𝐼 are the degrees of Truth, Fulsty, and Indetermincy
respective.

1.2.13 Neutrosophic Set

Let 𝑅 be a set of points with a general element 𝑣 denoted by 𝑥. The
neutrosophic set α is characterized by a true membership function 𝑇𝛼 , a

16
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

false membership function 𝐹𝛼 , and an undefined membership function 𝐼𝛼 .

These functions are real standard or non-standard subsets of the interval
[0,1], i.e. [25]:
𝑇𝛼 : 𝑣 → [0,1] 𝐹𝛼 : 𝑣 → [0,1] 𝐼𝛼 : 𝑣 → [0,1]
1.2.14 Neutrosophic Function
Functions used to transform nonlinear values into neutrosophic values
are called neutrosophic functions. For the function (𝑓𝑢𝑛), we distinguish
between two types of neutrosophic logic as follows [25]:
1. 𝑓𝑢𝑛0 : the function values start from zero (from bottom to top).
2. 𝑓𝑢𝑛1 : the function values start from one (from top to bottom).
The truth, indeterminacy, and falsity functions take their forms from
functions proposed independently of each other and do not necessarily
have the same form.
1.2.15 Neutrosophic Sample Space
The set 𝑋 of outputs of a random experiment is called a sample space.
When these outcomes include the outcome of our arrival at the non-
deterministic vector, we call 𝑋 the neutrosophic sample space [25].
1.2.16 Neutrosophic Event
It is a subset A of the sample space 𝑋 in a set of experimental results.
The neutrosophic sets of the sample space 𝑋 that are modified by various
for example combinations (which may include the undefined product of
the experimental results are called neutrosophic events [25].
1.2.17 Neutrosophic Random Variable
Let 𝛺 be the sample space of the neutrosophic random variable and
the neutrosophic random variable 𝑋𝑁𝑒 be a function of the sample space
𝛺. Thus, the sample space of neutrosophic random variables is the entire
possible space of a random experiment when it contains indeterminate

17
 CHAPTER ONE SOME BASIC CONCEPTS & RELATED DEFINITIONS

outcomes. That is, the true random variable with a real value, which is
defined as follows [25]:
𝑋: Ω → ℝ (1.11)
The neutrosophic random variable is defined as follows:
𝑋𝑁𝑒 : Ω → ℝ(I) (1.12)
Therefore, the form of the random variable 𝑋𝑁𝑒 = 𝑋 + 𝐼, where 𝑋𝑁𝑒 is
the neutrosophic random variable and 𝐼 is the uncertainty value.
1.2.18 The Probability Function and The Cumulative Distribution
Function of The Neutrosophic Random Variable
Suppose the neutrosophic random variable is 𝑋𝑁𝑒 = 𝑋 + 𝐼, then the
cumulative function (CDF) of the neutrosophic random variable,
abbreviated as NeCDF, can be found as 𝐹𝑋 (𝑥) = 𝑃(𝑋 ≤ 𝑥), that is [25]:
𝐹𝑋𝑁𝑒 = 𝐹𝑋 (𝑥 − 𝐼 ) (1.13)
𝑓𝑋𝑁𝑒 = 𝑓𝑋 (𝑥 − 𝐼 ) (1.14)
Since
𝐹𝑋 (𝑥 ) = 𝑃 (𝑋𝑁𝑒 ≤ 𝑥 )
𝐹𝑋 (𝑥 ) = 𝑃 (𝑋 + 𝐼 ≤ 𝑥 )
𝐹𝑋 (𝑥 ) = 𝑃 (𝑋 ≤ 𝑥 − 𝐼 )
The neutrosophic PDF, abbreviated as NePDF, can also be found by
deriving equation (1.13) with respect to 𝑋 as follows:
𝜕𝐹𝑋𝑁𝑒 (𝑥 ) 𝜕𝐹𝑋 (𝑥 − 𝐼 ) 𝜕 (𝑥 − 𝐼 )
= .
𝜕𝑥 𝜕𝑥 𝜕𝑥
𝐹𝑋 (𝑥 ) = 𝑓𝑋 (𝑥 − 𝐼 )
𝐹𝑋 (𝑥 ) = 𝑓𝑋𝑁𝑒

18
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

Chapter Two
A New Odd Weibull-G Family with Some of its
Properties

2.1 Introduction
Developing new probability distributions is essential in statistical
modeling, especially when dealing with real-world data that exhibit
variations in skewness, Kurtoses, and tail behavior. This chapter
introduces a new family of continuous distributions known as New Odd
Weibull-G (NOWG) family, which combines the Weibull distribution
with a hybrid methodology to enhance its flexibility and applicability. By
applying the hybrid transformation, the proposed family provides
expanded shape and scale properties, allowing it to better adapt to diverse
datasets.

This chapter provides a detailed derivation of the basic properties of

NOWG. Basic distributional functions, such as the pdf, and CDF, which
define model probabilities and the distribution of possible values, are
analyzed. In addition, the hazard function, which plays a role in reliability
and survival analysis, is studied along with statistical moments. These
properties provide a deeper understanding of the behavior of the proposed
distribution and its potential applications in various fields, paving the way
for discussions of parameter estimation, modeling fitting, and practical
applications in subsequent chapters.

2.2 The New Odd Weibull-G (NOWG) Family

Many studies have tried to find new distributions. These distributions
aim to extend existing ones and provide a better fit or to real data. For
example, in economics, better data fit more data-fit means less risk. In

19
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

2002, Eugene presented a novel technique 2002 for creating families

using continuous distributions, utilizing what was used [1]:

𝑅(𝑥)

𝑀(𝑥 ) = ∫ 𝓊(𝑡)𝑑𝑡 (2.1)

𝑎

where the range of 𝑡 ∈ [0,1], 𝓊(𝑡) is the pdf of the beta distribution, and
𝑅 (𝑥 ) is the CDF of any random variable.

In 2013, Alzaatreh et al. proposed a new method for generating

families from continuous distributions. The transformer is the name of a
random variable applied to modify an additional random variable. The
hazard functions are related to the resultant family of distributions. Each
distribution produced corresponds to the weighted hazard function of a
random variable weighted [2] by using the form:

𝒲(𝑅(𝑥,𝜙))

𝑀(𝑥, 𝜙) = ∫ 𝓊(𝑡)𝑑𝑡 (2.2)

𝑎

where 𝓊 (𝑡) is any pdf, 𝑅 (𝑥, 𝜙) is the baseline distribution and

𝒲(𝑅 (𝑥, 𝜙)) is a function of CDF that must satisfy the following three
conditions:
i. 𝒲(𝑅 (𝑥, 𝜙)) ∈ [𝑎, 𝑏], −∞ < 𝑎 < 𝑏 < ∞
ii. 𝒲(𝑅 (𝑥, 𝜙)) satisfies the following three conditions: differentiable, non-
decreasing and, non-decreasing monotonically.
iii. 𝒲(𝑅 (𝑥, 𝜙)) → 𝑎, 𝑎𝑠 𝑥 → −∞, and 𝒲(𝑅 (𝑥, 𝜙)) → 𝑏, 𝑎𝑠 𝑥 → ∞.

The literature has introduced several new types of distribution. They

are based on some 𝒲(𝑅 (𝑥, 𝜙)) functions, and some members of the T-X
family that are given in a table below:

20
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

Table (2.1). Some individuals in the T-X method

𝓦(𝐑(𝐱, 𝛟)) Range of 𝐱 Individuals in the T-X family

𝑅 (𝑥, 𝜙) [0,1] Beta-G [1], Mc-G [31]

− 𝑙𝑜𝑔[1 − 𝑅(𝑥, 𝜙)] (0, ∞) Gamma G Type1 [32]
− 𝑙𝑜𝑔[𝑅(𝑥, 𝜙)] (0, ∞) Gamma G Type2 [33]
− 𝑙𝑜𝑔[1 − 𝑅𝑤 (𝑥, 𝜙)] (0, ∞) Exponentiated T [34]
𝑅 (𝑥, 𝜙)
𝑙𝑜𝑔 [ ] (−∞, ∞) Logistic-G [35]
1 − 𝑅(𝑥, 𝜙)

𝑙𝑜𝑔[− 𝑙𝑜𝑔[1 − 𝑅(𝑥, 𝜙)] ] (−∞, ∞) Logistic-X [36]

− 𝑙𝑜𝑔[1 − 𝑅(𝑥, 𝜙)]
(0, ∞) Weibull-X [37]
1 − 𝑅(𝑥, 𝜙)

The steps to derive new family of distribution is done by

multiplying two values of 𝒲(𝑅 (𝑥, 𝜙)) that are 𝑅 (𝑥, 𝜙) [1], [31] and
− 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)] [32] in the form:
𝒲(𝑅 (𝑥, 𝜙)) = − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙) (2.3)
So that the above equation fulfills the conditions in [2] as follows:
i. − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙) ∈ [𝑎, 𝑏], −∞ < 𝑎 < 𝑏 < ∞
ii. − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙) satisfies the following three conditions:
differentiable, non-decreasing monotonically
iii. − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙) → 0, 𝑎𝑠 𝑥 → −∞,
and − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙) → 1, 𝑎𝑠 𝑥 → ∞
This approach is a development of the method 𝒲(𝑅 (𝑥, 𝜙)) in [2]
because it combines two methods for finding families of continuous
distributions in a new hybrid method.
This is referred to as the new generalized family of New Odd Weibull-
G distributions (NOWG). The CDF and pdf of Weibull distribution are
given as follows, which are provided in the following form:
𝜃
𝒰(𝑥 ) = 1 − 𝑒 −𝜅𝑥 (2.4)

21
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝜃
𝓊(𝑥 ) = 𝜅𝜃𝑥 𝜃−1𝑒 −𝜅𝑥 (2.5)
The CDF function is derived as follows for NOWG family by
substituting equations (2.3) and (2.5) into (2.1):
− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]𝑅(𝑥,𝜙)
𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = ∫ 𝜅𝜃𝑡 𝜃−1𝑒 −𝜅𝑡 𝑑𝑡 (2.6)
0
𝑅(𝑥,𝜙) ]𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]
or
𝑅(𝑥,𝜙) ]𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)] (2.7)

To get the pdf function for NOWG family equation (2.6) is derived,
then the pdf has the form:
𝑅(𝑥,𝜙) ]𝜃
𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 𝜅𝜃 𝑟(𝑥, 𝜙) × 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]
𝑅(𝑥,𝜙)
.[ − log(1 − 𝑅 (𝑥, 𝜙))] (2.8)
1−𝑅(𝑥,𝜙)
𝜃−1
. [−𝑅 (𝑥, 𝜙). log(1 − 𝑅 (𝑥, 𝜙))]
2.3 Some Properties of NOWG Family
2.3.1 Proposition: Lehmann Type II–Based CDF Transformation
Property for the NOWG Family
Proposition 2.3.1
Let 𝑋 be any arbitrary random variable with CDF 𝑅(𝑥, 𝜙) and pdf
𝑟(𝑥, 𝜙). Let 𝑇 ∈ (𝑎, 𝑏) be a random variable with a pdf 𝓊(𝑡).
Furthermore, let our suggested connection function using CDF and pdf be
as follows:
𝒲(𝑅 (𝑥, 𝜙)) = − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙)

𝜕𝒲(𝑅 (𝑥, 𝜙)) 𝑅(𝑥, 𝜙)

= 𝑟(𝑥, 𝜙) [ − log(1 − 𝑅 (𝑥, 𝜙))]
𝜕𝑥 1 − 𝑅 (𝑥, 𝜙)
Then the CDF function for NOWG has the form:

22
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝑅(𝑥,𝜙) ]𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]
Proof: By employing the Lehmman II as the Transformed (T),
where The Lehmann Type II transformation (often written as Lehmann II
or Lehmann alternative of type II) is a way of generating new
distributions from an existing baseline distribution 𝐹 (𝑥 ) by raising its
survival function to a positive power 𝜆 > 0. Given a baseline CDF 𝐹 (𝑥 )
and survival function 𝐹̅ (𝑥 ) = 1 − 𝐹 (𝑥 ):
𝐹𝑇 (𝑥 ) = 1 − [1 − 𝐹 (𝑥 )]𝜆
where 𝜆 > 0.
This transformation is called Lehmann Type II because it modifies
the upper tail by stretching or compressing it through the parameter 𝜆,
and the NOWG family as the Transformer (𝑋), we obtain:
− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]𝑅(𝑥,𝜙)
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = ∫ 𝜅𝜃 𝑟(𝑡, 𝜙)𝜒𝜙 𝑑𝑡
0

where
𝑅(𝑡,𝜙)
𝜒=[ − log(1 − 𝑅 (𝑡, 𝜙))]
1−𝑅(𝑡,𝜙)

𝜃−1 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑡,𝜙)]𝑅(𝑡,𝜙) ]𝜃

𝜙 = [−𝑅 (𝑡, 𝜙). log(1 − 𝑅 (𝑡, 𝜙))] 𝑒
Let 𝑍 = − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙)
𝑅 (𝑥, 𝜙)
⟹ 𝑑𝑍 = 𝑟(𝑥, 𝜙) [ − log(1 − 𝑅 (𝑥, 𝜙))] 𝑑𝑡
1 − 𝑅 (𝑥, 𝜙)
𝑘
𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = ∫ 𝜅𝜃𝑍𝑚−1𝑒 −𝜅𝑍 𝑑𝑍
0
𝜃 𝜃
Let 𝑦 = 𝑒 −𝜅𝑍 ⟹ 𝑑𝑦 = 𝜅𝜃𝑍𝑚−1𝑒 −𝜅𝑍 𝑑𝑍
when 𝑡 = 0, 𝑦 = 0 and
when , 𝑡 = 𝑍, 𝑦 = − 𝑙𝑜𝑔[1 − 𝑅 (𝑥, 𝜙)]𝑅(𝑥,𝜙)
− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]𝑅(𝑥,𝜙)
𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = ∫ 𝜅𝜃𝑍𝑚−1𝑒 −𝜅𝑍 𝑑𝑍
0

23
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝑅(𝑥,𝜙) ]𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]
when 𝑥 → 0, then
𝑅(𝑥,𝜙) ]𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]
=1−1= 0
when 𝑥 → ∞, then
𝑅(𝑥,𝜙) ]𝜃
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝑒 −𝜅[− 𝑙𝑜𝑔[1−𝑅(𝑥,𝜙)]
=1−0= 1
The equation (2.7) is obtained by taking the derivative of
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) respect to 𝑥, resulting in the probability density
function of the NOWG family of distributions.
It is enough to demonstrate that:
∞
∫ 𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) 𝑑𝑥 = 1
0
∞
∫ 𝜅𝜃 𝑟(𝑥, 𝜙)𝜒𝜙 𝑑𝑥
0
𝜃
Let ℎ = 𝑒 −𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
when 𝑥 = 0 then ℎ = 1, and 𝑥 = ∞ then ℎ = 0
𝜕ℎ
= 𝜅𝜃 𝑟(𝑥, 𝜙)𝜒𝜙
𝜕𝑥
1−𝜃
𝜕ℎ[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
𝑑𝑥 = 𝑅(𝑥,𝜙) 𝜃
𝜅𝜃 𝑟(𝑥,𝜙)[1−𝑅(𝑥,𝜙)−log(1−𝑅(𝑥,𝜙))]𝑒 −𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]

1
∫ 𝑑ℎ = [ℎ]10 = 1 − 0 = 1
0

Therefore

24
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝑅(𝑥, 𝜙)
𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 𝜅𝜃 𝑟(𝑥, 𝜙) [ − log(1 − 𝑅(𝑥, 𝜙))]
1 − 𝑅 (𝑥, 𝜙)
𝜃−1
. [−𝑅 (𝑥, 𝜙). log(1 − 𝑅 (𝑥, 𝜙))]
𝜃
. 𝑒 −𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
is a pdf function for NOWG family.

2.3.2 Expansions of the CDF and pdf

We can expand the CDF of the NOWG family using equation (2.6) as
follows: using an exponential expansion in form:
∞
−𝑡
(−1)𝑖
𝑒 =∑ 𝑡𝑖
𝑖!
𝑖=0

To get:
𝜃
𝑒 −𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
∞
(−1)𝑖(𝜃+1) 𝑖 𝑖𝜃
=∑ 𝜅 𝑅 (𝑥, 𝜙)𝑖𝜃 (log(1 − 𝑅 (𝑥, 𝜙)))
𝑖!
𝑖=0

Also, by using logarithm expansion which has a form:

∞
(𝑙𝑜𝑔(1 − 𝑡))𝑘 = ∑(−1)𝑖 𝑑𝑖,𝑘 𝑡 𝑖+𝑘
𝑖=0
[𝑚(𝑘+1)−𝑖]
𝑑𝑘,𝑖 = 𝑖 −1 ∑𝑖𝑚=1 for 𝑖 ≥ 0 and 𝑑𝑘,0 = 1
𝑚+1
∞
𝑖𝜃 𝑗+𝑖𝜃
(𝑙𝑜𝑔(1 − 𝑅 (𝑥, 𝜙))) = ∑(−1) 𝑗 𝑑𝑖𝜃,𝑗 (𝑅 (𝑥, 𝜙))
𝑗=0

𝑗 [𝑠(𝑖𝜃+1)−𝑗]
where 𝑑𝑖𝜃,𝑗 = 𝑗 −1 ∑𝑠=1 for 𝑗 ≥ 0 and 𝑑𝑖𝜃,0 = 1
𝑠+1

Then the final form for CDF of the NOWG family is:
𝑀𝐻𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 1 − 𝐻𝑅 (𝑥, 𝜙) 𝑗+2𝑖𝜃 (2.9)
(−1)𝑖(𝜃+1)+𝑗
where 𝐻 = ∑∞
𝑖,𝑗=0 𝑑𝑖𝜃,𝑗 𝜅 𝑖
𝑖!

To expand the function 𝑀𝐻𝑊 (𝑥, 𝜅, 𝜃, 𝜙)𝑡 which has a form:

25
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝜃 𝑡
𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙 )𝑡 = (1 − 𝑒 −𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
) (2.10)

𝜃 𝑡
−𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
Since 0 < (1 − 𝑒 ) < 1 we uses the

binomial series expansion as follows:

𝜃 𝑡
(1 − 𝑒 −𝜅[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))] )
∞
𝑡 𝜃
= ∑(−1)𝑞 ( ) 𝑒 −𝜅𝑞[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
𝑞
𝑞=0

In the same way, the expansion of the CDF function is found through
the following steps:
𝜃
𝑒 −𝜅𝑞[−𝑅(𝑥,𝜙).log(1−𝑅(𝑥,𝜙))]
∞
(−1)𝑙(𝜃+1) 𝑙𝜃
=∑ 𝜅 𝑙 𝑞𝑙 𝑅 (𝑥, 𝜙)𝑙𝜃 (log(1 − 𝑅 (𝑥, 𝜙)))
𝑖!
𝑙=0
𝑙𝜃
Also, by using logarithm expansion of (log(1 − 𝑅 (𝑥, 𝜙))) by the
form:
∞
𝑙𝜃 𝑢+𝑙𝜃
(𝑙𝑜𝑔(1 − 𝑅 (𝑥, 𝜙))) = ∑ (−1)𝑢 𝑑𝑙𝜃,𝑢 (𝑅 (𝑥, 𝜙))
𝑢=0
[𝑠(𝑙𝜃+1)−𝑢]
where 𝑑𝑖𝜃,𝑢 = 𝑢−1 ∑𝑢𝑠=1 for 𝑢 ≥ 0 and 𝑑𝑙𝜃,0 = 1
𝑠+1

finally to get the 𝑀𝐻𝑊 (𝑥, 𝜅, 𝜃, 𝜙)𝑡 by form:

𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙)𝑡 = Θ𝑅 (𝑥, 𝜙)𝑢+2𝑙𝜃 (2.11)
(−1)𝑞+𝑙(𝜃+1)+𝑢 𝑡
where Θ = ∑∞ ∞
𝑞=0 ∑𝑞=0 (𝑞 ) 𝑑𝑙𝜃,𝑢 𝜅 𝑙 𝑞𝑙
𝑖!

Now, the pdf of the NOWG family is expanded using equation

(2.7) as follows: Using the same way to expand the CDF function yields:

26
CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙)
∞
(−1)𝑖+2𝜃−1
=∑ 𝑅 (𝑥, 𝜙)𝜃(𝑖+1)(log(1
𝑖! [1 − 𝑅 (𝑥, 𝜙)]
𝑖=0
𝜃(𝑖+1)−1
− 𝑅(𝑥, 𝜙))) 𝜅 𝑖+1𝜃 𝑟(𝑥, 𝜙)
∞
(−1)𝑖+2𝜃−1
−∑ 𝑅(𝑥, 𝜙)𝜃(𝑖+1)−1(log(1
𝑖!
𝑖=0
𝜃(𝑖+1)
− 𝑅(𝑥, 𝜙))) 𝜅 𝑖+1𝜃 𝑟(𝑥, 𝜙)
𝜃(𝑖+1)
By expanding (log(1 − 𝑅 (𝑥, 𝜙))) using form:
∞
𝜃(𝑖+1) 𝑗+𝜃(𝑖+1)
(log(1 − 𝑅 (𝑥, 𝜙))) = ∑(−1)𝑗 𝑑𝜃(𝑖+1),𝑗 (𝑅 (𝑥, 𝜙))
𝑗=0
𝜃(𝑖+1)
Also expansion (log(1 − 𝑅 (𝑥, 𝜙))) by form:
𝜃(𝑖+1)−1
(log(1 − 𝑅 (𝑥, 𝜙)))
∞
𝑙+𝜃(𝑖+1)−1
= ∑(−1)𝑙 𝑑𝜃(𝑖+1)−1,𝑙 (𝑅 (𝑥, 𝜙))
𝑙=0
1
And using an expansion for to get a form:
1−𝑅(𝑥,𝜙)
∞
1
= ∑ 𝑅 (𝑥, 𝜙) 𝑧
1 − 𝑅 (𝑥, 𝜙)
𝑧=0

Finally, we get the expanded pdf for the NOWG Family in the form:
𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) = 𝑁𝑅 (𝑥, 𝜙)𝑙+2𝜃(𝑖+1)−1 𝑟(𝑥, 𝜙)
(2.12)
−𝐵𝑅 (𝑥, 𝜙)𝑙+2𝜃(𝑖+1)𝑟(𝑥, 𝜙)
where
∞
(−1)𝑖+2𝜃+𝑙−1
𝑁= ∑ 𝑑𝜃(𝑖+1)−1,𝑙 𝜅 𝑖+1𝜃
𝑖!
𝑖,𝑙,𝑧=0

And

27
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

∞
(−1)𝑖+2𝜃+𝑗−1
𝐵= ∑ 𝑑𝜃(𝑖+1),𝑗 𝜅 𝑖+1𝜃
𝑖!
𝑖,𝑗=0
[𝑠(𝜃(𝑖+1)−1+1)−𝑙]
where 𝑑𝜃(𝑖+1)−1,𝑙 = 𝑙−1 ∑𝑙𝑠=1 for 𝑙 ≥ 0 and 𝑑𝜃(𝑖+1)−1,0 =
𝑠+1
𝑗 [𝑙(𝑖𝜃(𝑖+1)+1)−𝑗]
1, and 𝑑𝜃(𝑖+1),𝑗 = 𝑗 −1 ∑𝑙=1 for 𝑗 ≥ 0 and 𝑑𝜃(𝑖+1),0 = 1
𝑙+1

2.3.3 Quantile Function of NOWG Family

The quantile function is known as the inverse of the cumulative
distribution function and is used to find the median, skewness, and
kurtosis of distributions with large skewed values or that do not contain
moments. It can be used to generate random numbers for data for
simulation study and defined by the form [4], [38]:
𝑄 (𝑢) = 𝐹−1(𝑢)
where 𝑄 (𝑢) is the quantile function 𝑀𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) for each
𝑢~𝑈[0,1]. We find the quantile function for the NOWG family by
assuming 𝑍 = 𝑅 (𝑥, 𝜙), then:
𝜃
𝑢 = 1 − 𝑒 −𝜅[−𝑍.log(1−𝑍)]
𝜃
Then 1 − 𝑢 = 𝑒 −𝜅[−𝑍.log(1−𝑍)]
log(1−𝑢)
Then − = [−𝑍. log(1 − 𝑍)]𝜃
𝜅
1
log(1−𝑢) 𝜃
Then 𝑍. log(1 − 𝑍) = − (− )
𝜅
1
log(1−𝑢) 𝜃
Putting 𝛽 = − (− )
𝜅

Then 𝑍. log(1 − 𝑍) = 𝛽, hence:

𝛽
log(1 − 𝑍) = … (∗)
𝑍
𝛽
Let 𝜉 =
𝑍
𝛽
𝑍=
𝜉

28
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝛽 𝜉−𝛽
1−𝑍 = 1− =
𝜉 𝜉
Substituting into the equation (∗), we get:
𝜉−𝛽
𝑙𝑜𝑔 ( )=𝜉
𝜉
𝜉−𝛽
= 𝑒𝜉
𝜉
𝜉−𝛽
𝑒 −𝜉 =1
𝜉
𝑥−𝑝
[From Theorem (the solution (𝑠) of the equation 𝑒 𝑎𝑥 . = 𝑏 are
𝑥−𝑞

1 𝑞
𝑥 = 𝑝 + . 𝑊− 𝑏𝑒 −𝑎𝑝 (𝑎𝑏𝑒 −𝑎𝑝 𝑇), 𝑊 (𝑝; 𝑏) = 𝑝 + 𝑊− 𝑏𝑒 −𝑝 (𝑏𝑒 −𝑝 𝑇),
𝑎

𝑇 = 𝑝 − 𝑞)]
Where from the above Theorem rut 𝑎 = −1, 𝑝 = 𝛽, 𝑞 = 0, 𝑏 = 1,
we have:
𝜃 = 𝛽 + 𝑊−1(𝛽𝑒 −𝛽 )
𝛽
= 𝛽 + 𝑊−1(𝛽𝑒 −𝛽 )
𝑍
𝛽
𝑍=
𝛽 + 𝑊−1(𝛽𝑒 −𝛽 )
Hence we get:
𝛽
𝑅 (𝑥, 𝜙) =
𝛽 + 𝑊−1(𝛽𝑒 −𝛽 )
𝛽
𝑄𝑀𝑁𝑂𝑊 (𝑥,𝜅,𝜃,𝜙) = (2.13)
𝛽 + 𝑊−1(𝛽𝑒 −𝛽 )
1
log(1−𝑢) 𝜃
Where 𝛽 = − (− ) , and 𝑊−1(𝛽𝑒 −𝛽 ) is the lower Lombart
𝜅

function for 𝛽𝑒 −𝛽 ∈ [𝑒 −1, 0).

29
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

2.3.4 Moments of NOWG Family

Moments play a crucial role in determining the mean, variance,
skewness, and flatness of a probability distribution. Let 𝑥 be a random
variable with pdf for NOWG family. Then the 𝑛𝑡ℎ moment of any
distribution is given by [39]:
∞

μ𝑛 = 𝐸 (𝑥 𝑛 )𝑁𝑂𝑊𝐺 = ∫ 𝑥 𝑛 𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙)𝑑𝑥

0

By substitution, the pdf 𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙), of NOWG Family has a

form:
∞

μ𝑛 = 𝑁 ∫ 𝑥 𝑛 𝑅(𝑥, 𝜙)𝑙+2𝜃(𝑖+1)+𝑧−1 𝑟(𝑥, 𝜙)𝑑𝑥

0
∞
(2.14)

−𝐵 ∫ 𝑥 𝑛 𝑅 (𝑥, 𝜙)𝑙+2𝜃(𝑖+1)𝑟(𝑥, 𝜙)𝑑𝑥

0

2.3.5 Moment Generating Function of NOWG Family

The moment generating function (𝑚𝑔𝑓) is used to derive moments for
distributions whose moments are difficult to be found directly, determine
the distribution of sums and variances of independent random variables,
and derive finite distributions for sequences of random variables. This
function is an integral transformation of the density function and can be
used to calculate moments about zero and central moments. It also
calculates various moments of a distribution, such as the mean (first
moment), variance (second moment), and others. It also analyzes the
statistical behavior of random variables and facilitates complex
mathematical operations related to statistical distributions. The moment
generating function (𝑚𝑔𝑓) is given by [40]:

30
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

𝑀𝑥 (𝑡)𝑁𝑂𝑊𝐺 = 𝐸 (𝑒 𝑡𝑥 ) = ∫ 𝑒 𝑡𝑥 𝑚𝐻𝑊 (𝑥, 𝜅, 𝜃, 𝜙)𝑑𝑥

−∞

Using series expansion for 𝑒 𝑡𝑥

∞ 𝑡𝑛 𝑛)
∞ 𝑦𝑛
𝑀𝑥 (𝑦)𝑁𝑂𝑊𝐺 = ∑ (
𝐸 𝑥 =∑ [μ𝑛 ]
𝑛=0 𝑛! 𝑛=0 𝑛!

From equation (2.14), we get:

∞
𝑡𝑛
∞
μ𝑛 = ∑ [𝑁 ∫ 𝑥 𝑛 𝑅 (𝑥, 𝜙)𝑙+2𝜃(𝑖+1)+𝑧−1 𝑟(𝑥, 𝜙)𝑑𝑥
𝑛=0 𝑛!
0
∞
(2.15)

− 𝐵 ∫ 𝑥 𝑛 𝑅 (𝑥, 𝜙)𝑙+2𝜃(𝑖+1)𝑟(𝑥, 𝜙)𝑑𝑥 ]

0

2.3.6 Probability Weighted Moments

The probability-weighted moments of the NOWG family can be found
using the following equation [41]:
∞

𝜏ℎ,𝑡 = 𝐸 (𝑥 ℎ 𝐹𝑡 (𝑋)) = ∫ 𝑥 ℎ 𝑓 (𝑥 )𝐹𝑡 (𝑥 )𝑑𝑥

−∞

Substituting 𝑀𝑁𝑂𝑊 (𝑥, 𝑤, 𝑚, 𝜂)𝑡 and 𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) for the NOWG
Family into the previous equation, we get:
∞

𝜏ℎ,𝑠 = ∫ 𝑥 ℎ Θ𝑅 (𝑥, 𝜂)𝑢+2𝑙𝑚[𝑁𝑅 (𝑥, 𝜂)𝑘+2𝑖𝑚+2𝑚+𝑧−1𝑟(𝑥, 𝜂)

0
(2.16)

− 𝑀𝑅 (𝑥, 𝜂)𝑘+2𝑖𝑚+2𝑚 𝑟(𝑥, 𝜂)]𝑑𝑥

Finally
∞

𝜏ℎ,𝑠 = ∫ 𝑥 ℎ 𝑁𝑅 (𝑥, 𝜂)𝑢+2𝑙𝑚+𝑘+2𝑖𝑚+2𝑚+𝑧−1𝑑(𝑥, 𝜂)𝑑𝑥

0
∞
(2.17)

− ∫ 𝑥 ℎ Θ𝑁𝑅 (𝑥, 𝜂)𝑢+2𝑙𝑚+𝑘+2𝑖𝑚+2𝑚𝑟(𝑥, 𝜂)𝑑𝑥

0

31
 CHAPTER TWO A NEW ODD WEIBULL-G FAMILY WITH SOME OF ITS PROPERTIES

2.3.7 Incomplete Moments

The incomplete moments of the NOWG family can be found using the
following equation [41]:
𝑦

𝜇𝑟 (𝑦) = ∫ 𝑥 𝑟 𝑓(𝑥 )𝑑𝑥

0

Substituting 𝑚𝑁𝑂𝑊 (𝑥, 𝜅, 𝜃, 𝜙) for the NOWG family into the previous
equation, we get:
𝑦

𝜇𝑟 (𝑦) = ∫ 𝑥 𝑟 [𝑁𝑅 (𝑥, 𝜂)𝑘+2𝑖𝑚+2𝑚+𝑧−1 𝑟(𝑥, 𝜂)

(2.18)
0

− 𝑀𝑅 (𝑥, 𝜂)𝑘+2𝑖𝑚+2𝑚 𝑟(𝑥, 𝜂)]𝑑𝑥

2.3.8 Rényi Entropy

The Rényi entropy for the NOWG family distribution can be obtained
as [42]:
∞
1
𝐼𝑐 (𝑥 )𝐻𝑊𝐺 = 𝑙𝑜𝑔 ∫ 𝑓(𝑥 )𝑐 𝑑𝑥
1−𝑐
0

Then from equation (2.12), we get:

∞
1
𝐼𝑐 (𝑥 )𝐻𝑊𝐺 = log [∫ (𝑁𝑅 (𝑥, 𝜂)𝑘+2𝑖𝑚+2𝑚+𝑧−1 𝑟(𝑥, 𝜂)
1−𝑐
0

𝑐
− 𝑀𝑅 (𝑥, 𝜂)𝑘+2𝑖𝑚+2𝑚 𝑟(𝑥, 𝜂)) 𝑑𝑥]

Finally, we get:
𝑤
∞ 𝑅 (𝑥, 𝜂 )
log [∑𝑐𝑣=0(−1)𝑣 (𝑣𝑐 )𝑁𝑀 ∫0 𝑑𝑥]
𝑟(𝑥, 𝜂)−𝑐 (2.19)
𝐼𝑐 (𝑥 )𝐻𝑊𝐺 =
1−𝑐

where 𝑤 = 𝑣(𝑘 + 2𝑖𝑚 + 2𝑚) + 𝑣(𝑧 − 1)

32
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

Chapter Three

A New Odd Weibull Inverse Burr Type X, Properties

with Application

3.1 Introduction
In this chapter, the Burr type X distribution is represented as a mixture
with the New odd Weibull family, called New Odd Weibull Inverse Burr
Type X (NOWIBX) distribution to demonstrate its usefulness in
modeling statistical patterns in extreme values recorded in non-stationary
streams of media events. The basic functions of the new distribution are
presented in addition to several statistical and mathematical properties of
the hybrid distribution as well as estimating the model parameters by five
methods and conduct a Monte Carlo simulations for the five methods.
Finally, these results are illustrated through an example of matching the
hybrid distribution with data representing the values of Bladder cancer
and survival times of a group of 46 patients (in years) treated with
chemotherapy alone and knowing the results of the improvement of this
expansion through comparison with some other distributions using some
statistical metrics.

3.2 New Odd Weibull Inverse Burr-X (NOWIBX)

Distribution
Let 𝑋 represent the random variable, the CDF and pdf functions of
Inverse Burr X with two-parameter respectively, are [43]:
𝛾
𝛼2
−
𝐺 (𝑥, 𝛼, 𝛾) = 1 − (1 − 𝑒 𝑥2 ) , 𝑥 ≥ 0, 𝛼, 𝛾 > 0 (3.1)

𝛼2 𝛼2 𝛾−1
2 −3 − 𝑥 2 −
(3.2)
𝑔(𝑥, 𝛼, 𝛾) = 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥2 )

33
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

By substituting equation (3.1) into equation (2.7), we obtain the CDF

function for the new distribution (New Odd Weibull Inverse Burr-X
(NOWIBX) distribution) in the form:
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ] (3.3)
𝐹 (𝑥 ) = 1 − 𝑒
where , 𝑥 ≥ 0 and 𝛿, 𝜃, 𝛼, 𝛾 > 0 are shape parameters of NOWIBX.

Using programming in the R software, the CDF function is drawn for

the NOWIBX distribution for different parameter values, as shown in
Figure (3.1).

Figure (3.1). CDF function of NOWIBX distribution for different parameter values

To find the pdf function for the NOWIBX distribution, equation (3.3)
is derived or equations (3.1) and (3.2) are substituted into equation (2.8)
to obtain the equation in the form:

34
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝛼2 𝛼2 𝛾−1
2 −3 − 𝑥 2 −
𝑥2 )
𝛿𝜃 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒
𝑓 (𝑥 ) = 𝛾 𝛾 1−𝜃
𝛼2 𝛼2
− 2 − 2
[− (1 − (1 − 𝑒 𝑥 ) ) 𝑙𝑜𝑔 (1 −𝑒 𝑥 ) ]

𝛼2 𝛾
− (3.4)
1 − (1 − 𝑒 𝑥2 ) 𝛾
𝛼2
− 2
. − log (1 − 𝑒 𝑥 )
𝛼2 𝛾
−
(1 − 𝑒 𝑥2 )
[ ]
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ]
.𝑒

Using programming in the R software, the pdf function is drawn for

the NOWIBX distribution for different parameter values, as shown in
Figure (3.2).

Figure (3.2). pdf function of NOWIBX distribution for different parameter values

The Survival function has the form:

𝑆(𝑥 )𝑁𝑂𝑊𝐼𝐵𝑋 = 1 − 𝐹 (𝑥 ) (3.5)
Thus, by substituting equation (3.3) into the aforementioned equation,
we obtain:

35
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ] (3.6)
𝑆(𝑥 )𝑁𝑂𝑊𝐼𝐵𝑋 = 𝑒
Using programming in the R software, the survival function is drawn
for the NOWIBX distribution for different parameter values, as shown in
Figure (3.3).

Figure (3.3). Survival function of NOWIBX distribution for different parameter

values

The Hazard function, which holds significant importance, particularly

about matters of life, has garnered considerable interest from researchers.
They have concentrated their efforts on identifying statistical distributions
of various kinds for this function. Consequently, the following
relationship can be utilized to derive it using equation:
𝑓(𝑥 )
ℎ(𝑥 )𝑁𝑂𝑊𝐼𝐵𝑋 = (3.7)
𝑆(𝑥 )𝑁𝑂𝑊𝐼𝐵𝑋
Thus, by substituting equation (3.4) and (3.7) into the previous
equation, we obtain:

36
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝛾
𝛼2
− 2
𝛾−1 1−(1−𝑒 𝑥 ) 𝛾
𝛼2 𝛼2 𝛼2
2 −3 − 2 − 2 − 2
𝛿𝜃 2𝛼 𝛾𝑥 𝑒 𝑥 (1−𝑒 𝑥 ) 𝛾 −log(1−𝑒 𝑥 )
𝛼2
− 2
(1−𝑒 𝑥 ) (3.8)
[ ]
ℎ(𝑥 ) = 𝛾 𝛾 1−𝜃
𝛼2 𝛼2
− 2 − 2
[−(1−(1−𝑒 𝑥 ) )𝑙𝑜𝑔(1−𝑒 𝑥 ) ]

Using programming in the R software, the hazard function is drawn

for the NOWIBX distribution for different parameter values, as shown in
Figure (3.4).

Figure (3.4). Hazard function of NOWIBX distribution for different parameter values

Figure 1 shows the CDF of the NOWIBX distribution for different

parameter values. The graph shows that all curves start at zero and
gradually approach one as the variable values increase, confirming that
the distribution satisfies the cumulative distribution function property.
Varying parameters alter the shape of the curve; some values cause the
distribution to rise rapidly toward one (a more centered distribution),
while other values cause it to rise more slowly (a more dispersed

37
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

distribution). This demonstrates the flexibility of the distribution to

represent a variety of data.

Figure 2 shows the pdf of the NOWIBX distribution for the same set
of different parameters. The graph shows that the shape of the curve is
clearly affected by the parameters, as it can be symmetrical or skewed,
sometimes with sharp or flat peaks. These variations in the shape of the
pdf indicate the distribution's ability to represent data with different
characteristics, such as positive or negative skewness and high or low
centering, which gives it an advantage in practical applications.

Figure 3 shows the survival function for the same distribution. As

expected, the curves start at 1 and gradually decrease toward zero as the
variable increases, with different rates of decrease depending on the
parameter values. This behavior gives the distribution flexibility in
modeling lifetime or failure data in reliability systems, as it can represent
cases of rapid or slow declines in survival probabilities.

Figure 4 shows the hazard function, which represents the

instantaneous risk of failure. The resulting shapes in this graph vary,
showing patterns similar to an "inverted bathtub" or increasing or
decreasing patterns, depending on the parameter values. This diversity
reflects the distribution's ability to represent systems where the hazard
rate varies over time, an important property in fields such as survival and
reliability analysis.

3.3 Statistical Properties of NOWIBX Distribution

3.3.1 Expansions CDF, CDFs, and pdf
In order to obtain expanded CDF and pdf functions, exponential and
logarithmic function expansions and binomial series expansions are used.
First, the CDF function is expanded as follows:

38
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ]
𝑒
∞
(−1)𝑖(1+𝜃) 𝑖
=∑ 𝛿 [(1
𝑖!
𝑖=0
𝛾 𝛾 𝑖𝜃
𝛼2 𝛼2
− 2 −
− (1 − 𝑒 𝑥 ) ) log (1 − 𝑒 𝑥2 ) ]

𝛼2
𝛾 𝑖𝜃
−
By using logarithm expansion for (𝑙𝑜𝑔 (1 − 𝑒 𝑥2 ) ) by the form:

𝑖𝜃
𝛼2 𝛾
− 2
(log [1 − 1 + (1 − 𝑒 𝑥 ) ])

∞ 𝑗+𝑖𝜃
𝛼2 𝛾
−
= ∑(−1)𝑗 𝐷𝑖𝜃,𝑗 (1 − (1 − 𝑒 𝑥2 ) )
𝑗=0
𝑗 𝑚(𝑖𝜃+1)−𝑗
where 𝐷𝑖𝜃,𝑗 = 𝑗 −1 ∑𝑚=1 𝐷𝑖𝜃,𝑗−𝑚 , 𝑗 ≥ 1, 𝐷𝑖𝜃,0 = 1
𝑚+1

∞ 𝛾 𝑗+2𝑖𝜃
(−1)𝑖(1+𝜃)+𝑗 𝑖
𝛼2
− 2
∴ 𝐹 (𝑥 ) = 1 − ∑ 𝐷𝑖𝜃,𝑗 𝛿 (1 − (1 − 𝑒 𝑥 ) )
𝑖!
𝑖,𝑗=0
𝛼2
𝛾 𝑗+2𝑖𝜃
−
Now by binomial series expansion for (1 − (1 − 𝑒 𝑥2 ) )

we get:
𝑗+2𝑖𝜃 ∞
𝛼2 𝛾 𝑗 + 2𝑖𝜃 𝛼2 𝑘𝛾
− 𝑘 − 2
(1 − (1 − 𝑒 𝑥2 ) ) = ∑ (−1) ( ) (1 − 𝑒 𝑥 )
𝑘
𝑘=0
𝛼2
𝑘𝛾
−
Again by using binomial series expansion for (1 − 𝑒 𝑥2 ) we get:
∞
𝛼2 𝑘𝛾 𝑘𝛾 −𝑝−𝛼22
−
(1 − 𝑒 𝑥2 ) = ∑ (−1 )𝑝 ( )𝑒 𝑥
𝑝
𝑝=0
Finally, the CDF of NOWIBX distribution have the form:
𝛼2
−𝑝− (3.9)
𝐹 (𝑥 ) = 1 − Ω𝑒 𝑥2

(−1)𝑖(1+𝜃)+𝑗+𝑘+𝑝
where Ω = ∑∞
𝑖,𝑗,𝑘,𝑝=0 𝐷𝑖𝜃,𝑗 (𝑗+2𝑖𝜃
𝑘
) (𝑘𝛾
𝑝
) 𝛿𝑖
𝑖!

39
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

While the CDFs has form:

𝛾 𝛾 𝜃
𝑠
𝛼2 𝛼2
− 2 − 2
−𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ]
𝑑(
𝐹 𝑥) = 1−𝑒

( )
By using binomial series expansion, get the form:
𝑠
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ]
1−𝑒

( )
𝛾 𝛾 𝜃
𝛼2 𝛼2
∞ − 2 − 2
−𝛿𝑙[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ]
𝑠
= ∑(−1)𝑙 ( ) 𝑒
𝑙
𝑙=0
By same steps used for the CDFs, we get:

𝛼2
𝑠( −𝑞− (3.10)
𝐹 𝑥 ) = 𝛫𝑒 𝑥2

(−1)𝑙+𝑣+𝑢(1+𝑙𝜃)+𝑠+𝑞 𝑑 𝑣+2𝑢𝜃𝑙
where Κ = ∑∞
𝑙,𝑢,𝑣,𝑠,𝑞=0 ( 𝑙 )( 𝑠 ) (𝑠𝛾𝑞
) 𝛿 𝑢 𝑙𝑢 𝐷𝑢𝜃𝑙,𝑣 ,
𝑢!
𝑗 𝑚(𝑢𝜃𝑙+1)−𝑣
and 𝐷𝑢𝜃𝑙,𝑣 = 𝑣 −1 ∑𝑚=1 𝐷𝑢𝜃𝑙,𝑣−𝑚 , 𝑗 ≥ 1, 𝐷𝑢𝜃𝑙,0 = 1
𝑚+1
To expand the pdf of NOWIBX distribution it is done and has
following form:
𝛼2 𝛼2 𝛾(ℎ+𝑠)−1
2 −3 − 𝑥 2 −
𝑓 (𝑥 ) = Θ2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥2 )

(3.11)
𝛼2 𝛼2 𝛾(𝑠+1)−1
2 −3 −(𝑤+𝛾+1)𝑥 2 − 2
− Β 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥 )

where
(−1)𝑖(𝜃+1)+𝜃−1+𝑧+𝑠+ℎ
Θ = ∑∞
𝑖,𝑧,𝑠,ℎ=0 𝛿 𝑖+1𝐷𝜃(𝑖+1)−1,𝑧 (𝑧𝑠)(ℎ1) 𝜃
𝑖!
(−1)𝑖(𝜃+1)+𝜃−1+𝑧+𝑠+𝑤
Β = ∑∞
𝑖,𝑧,𝑠,𝑤=0 𝛿 𝑖+1𝐷𝛾,𝑤 𝐷𝜃(𝑖+1)−1,𝑧 (𝑧𝑠) 𝜃
𝑖!
𝑐(𝛾+1)−𝑤
𝐷𝛾,𝑤 = 𝑤 −1 ∑𝑙𝐶=1 𝐷𝛾,𝑤−𝑐 , 𝛾 ≥ 1, 𝐷𝛾,0 = 1
𝑐+1

40
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝑚(𝜃(𝑖+1))−𝑧
𝐷𝜃(𝑖+1)−1,𝑧 = 𝑧 −1 ∑𝑙𝑚=1 𝐷𝜃(𝑖+1)−1,𝑧−𝑚 , 𝑧 ≥ 1, 𝐷𝜃(𝑖+1)−1,0 =
𝑧+1
1

3.3.2 Quantile Function of NOWIBX Distribution

The quantile function 𝑄(𝑢) is derived from equation (2.13) have
get:

𝛼2
𝑥= − 1
√ 𝜌 𝛾 (3.12)
log (1 − [1 − ] )
𝜌 − 𝑊−1 −𝜌 exp[𝜌])
(

1
log(1−𝑢) 𝜃
where 𝜌 = − (− ) , and 𝑊−1 (𝜌𝑒 −𝜌 ) is lower Lombart
𝛿

function for 𝜌𝑒 −𝜌 .
The following table represents the quantile function values for
different values of parameters.

Table(3.1). The quantiles for some parameter values of NOWIBX distribution.

(𝛿, 𝜃, 𝛼, 𝛾)
𝒖
(1.3,0.7,0.2,0.4) (1.4,0.3,0.7,0.3) (0.5,0.8,0.3,0.7) (1.7,3,0.8,0.3) (1.6,0.3,0.1,0.2)
𝟎. 𝟏 0.1957 0.3959 0.3355 2.7336 0.0582
𝟎. 𝟐 0.2527 0.5048 0.4488 3.3793 0.0754
𝟎. 𝟑 0.3151 0.6340 0.5842 3.9337 0.0965
𝟎. 𝟒 0.3927 0.8157 0.7716 4.4723 0.1272
𝟎. 𝟓 0.4989 1.1144 1.0652 5.0343 0.1797
𝟎. 𝟔 0.6610 1.7215 1.5997 5.6568 0.2910
𝟎. 𝟕 0.9484 3.5055 2.8097 6.3949 0.6331
𝟎. 𝟖 1.6086 16.0204 6.7854 7.3610 3.0569
𝟎. 𝟗 4.3913 4632.5812 37.3498 8.9023 589.5359

Table (3.1) displays the values of the quantile function for different
sets of parameters. These values represent the points that divide the
distribution into equally likely parts. The key observation is that changing
the parameters leads to a marked difference in the quantile values for the
same ratio. Some sets produce very small quantiles, while others produce
very large values. This reflects the distribution's sensitivity to shape

41
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

parameters, enabling it to adapt to data with different ranges and unequal

distributions. The regular increase in quantile values with the ratio
confirms the validity of the distribution's behavior and its consistency
with the theoretical properties of quantile functions.

3.3.3 Moments
The 𝑟 𝑡ℎ moment of NOWIBX distribution can be finding from
equation (2.14) and (3.11) we get a form:

∞
𝛼2 𝛼2 𝛾(ℎ+𝑠)−1
𝑟 2 −3 − 𝑥 2 −
𝑥2 )
μ𝑟 = ∫ 𝑥 [Θ2𝛼 𝛾𝑥 𝑒 (1 − 𝑒
0

𝛼2 𝛼2 𝛾(𝑠+1)−1
2 −3 −(𝑤+𝛾+1)𝑥 2 −
𝑥2 )
− Β2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 ] 𝑑𝑥

∞ 𝛾(ℎ+𝑠)−1
𝛼2 𝛼2
2 𝑟−3 − 𝑥 2 − 2
= Θ ∫ 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥 ) 𝑑𝑥
0
∞
𝛼2 𝛼2 𝛾(𝑠+1)−1
2 𝑟−3 −(𝑤+𝛾+1)𝑥 2 −
𝑥2 )
− Β∫ 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑑𝑥
0
1
By using the change of variable, 𝑡 = 𝛼2
, 0 < 𝑡 < 1, we obtain:
−
1−𝑒 𝑥2
1 𝑟
1
(ℎ+𝑠)−1 1 2
μ 𝑟 = Θ ∫ (𝛼 2 𝛾 ) 𝑡 (− [ln ( − 1)] + 1) 𝑑𝑡
0 𝑡

𝑟
1
1 1 2
𝛾(𝑠+1)−1
−Β ∫ (𝛼 2𝛾)𝑡 (− [ln ( − 1)] + (𝑤 + 𝛾 + 1)) 𝑑𝑡
0 𝑡

42
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝑟 𝑖
1 1 2
𝑟 (ℎ+𝑠)−1
μ𝑟 = Θ𝛼 𝛾 ∑ ( ) (−1) ∫ (𝛼 2 𝛾)𝑡
2 𝑟
[ln ( − 1)] 𝑑𝑡
𝑖 0 𝑡
𝑖=0
𝑟
𝑟
− Β𝛼 2 𝛾 ∑ ( 𝑗 ) (−1)𝑟 (𝑤 + 𝛾
𝑗=0
𝑗
1 1 2
𝛾(𝑠+1)−1
+ 1)𝑟−𝑗 ∫ (𝛼 2 𝛾)𝑡 [ln ( − 1)] 𝑑𝑡
0 𝑡
1
Now by using − 1 = 𝑒 −𝑘 , 0 < 𝑘 < ∞, we obtain
𝑡
𝑟 ∞
𝑟 𝑖
μ𝑟 = Θ𝛼 𝛾 ∑ ( ) (−1)𝑟 ∫ (𝑒 −𝑘 + 1)(ℎ+𝑟)−1𝑢2𝑒 𝑢 𝑑𝑢
2
𝑖 0
𝑖=0
𝑟 ∞
2
𝑟 𝑟( 𝑟−𝑗 −𝑘
𝑗
𝛾(𝑟+1)−1 2 𝑢
( )
−Β𝛼 𝛾 ∑ ( 𝑗 ) −1 𝑤 + 𝛾 + 1) (
∫ 𝑒 +1 ) 𝑢 𝑒 𝑑𝑢
𝑗=0 0

𝑟
𝑟
μ𝑟 = Θ𝛼 2𝛾 ∑ ( ) (−1)𝑟 𝐸𝑞 (𝑔(𝑋))
𝑖
𝑖=0
𝑟
(3.13)
𝑟
−Β𝛼 2 𝛾 ∑ ( 𝑗 ) (−1)𝑟 (𝑤 + 𝛾 + 1)𝑟−𝑗 𝐸𝑞 (𝑘 (𝑋))
𝑗=0

where 𝐸𝑞 (∗) is expressing to expectation for X~q, and 𝑞 is standard

𝑖
exponential distribution and 𝑔(𝑋) = (𝑒 𝑥 + 1)(ℎ+𝑟)−1𝑥 2 𝑒 2𝑥 , 𝑘 (𝑋) =
𝑗
(𝑒 𝑥 + 1)𝛾(𝑟+1)−1𝑥 2 𝑒 2𝑥
The importance sampling approach yields an estimate of 𝜇𝑟 , which is
referred to as the importance sampling estimate.
𝑟 𝑚
𝑟 1
𝜇̂ 𝑟,𝑞 = Θ𝛼 2𝛾 ∑ ( ) (−1)𝑟 ( ∑ 𝑔(𝑋𝑎 ))
𝑖 𝑚
𝑖=0 𝑎=0
𝑟 𝑚
(3.14)
𝑟 1
−Β𝛼 2𝛾 ∑ ( 𝑗 ) (−1)𝑟 (𝑤 + 𝛾 + 1)𝑟−𝑗 ( ∑ 𝑘 (𝑋𝑎 ))
𝑚
𝑗=0 𝑎=0

Mean of 𝜇̂ 𝑟,𝑞 is given by:

43
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝐸(𝜇̂ 𝑟,𝑞 ) = 𝜇𝑟 (3.15)

The variance 𝜇̂ 𝑟,𝑞 are given by:
𝑟
𝑟 2
𝑣𝑎𝑟(μ̂𝑟,𝑞 ) = Θ𝛼 𝛾 ∑ [( )] 𝑣𝑎𝑟 (𝐸̂𝑞 (𝑔(𝑋))) − 2ℋ
4 2
𝑖
𝑖=0
𝑟
(3.16)
𝑟 2
+Β𝛼 4𝛾 2 ∑ [( 𝑗 ) (𝑤 + 𝛾 + 1)𝑟−𝑗 ] 𝑣𝑎𝑟 (𝐸̂𝑞 (𝑘 (𝑋) ))
𝑗=0

where
𝑟 𝑟
ℋ = 𝑐𝑜𝑣 (Θ𝛼 2 𝛾 ∑𝑟𝑖=0 ( ) (−1)𝑟 𝐸𝑞 (𝑔(𝑋) ), Β𝛼 2 𝛾 ∑𝑟𝑗=0 ( 𝑗 ) (−1)𝑟 (𝑤 +
𝑖

𝛾 + 1)𝑟−𝑗 𝐸𝑞 (𝑘 (𝑋)))

2
𝑣𝑎𝑟 (𝐸̂𝑞 (𝑔(𝑋))) = 𝐸𝑞 (𝑔 (𝑋) − 𝐸𝑞 (𝑔(𝑋)) ),
2
and (𝐸̂𝑞 (𝑘 (𝑋))) = 𝐸𝑞 (𝑘 (𝑋) − 𝐸𝑞 (𝑘 (𝑋)) )

Table (3.2) represents the values of the first, second, third, and fourth
moments, in addition to the values of skewness and kurtoses for different
values of parameters.
Table (3.2): Numerical value of 𝜇1 , 𝜇2 , 𝜇3 , 𝜇4 , 𝜎 2 , 𝑆𝐾, and 𝐾𝑈 of the NOWIBX
distribution
𝛿 𝜃 𝛼 𝛾 𝜇1 𝜇2 𝜇3 𝜇4 𝜎2 𝑆𝐾 𝐾𝑈
1.1 0.46171 0.23710 0.13734 0.09147 0.02393 1.18955 1.62715
1.5 0.3
1.2 0.43477 0.20746 0.10979 0.06532 0.01844 1.16181 1.51773
1.2
1.3 0.55229 0.34287 0.24576 0.07654 0.03784 1.22410 0.65108
1.3 0.4
1.4 0.52724 0.30855 0.20460 0.15889 0.03056 1.19378 1.66893
1.5 0.55073 0.31642 0.18976 0.11885 0.01312 1.06609 1.18704
1.5 0.5
1.6 0.53464 0.29728 0.17196 0.10353 0.01144 1.06094 1.17149
2.2
1.7 0.63551 0.41628 0.28095 0.19529 0.01240 1.04602 1.12693
1.7 0.6
1.8 0.61997 0.39543 0.25937 0.17488 0.01106 1.04305 1.11838

Table (3.2) presents the numerical values of the first four moments,
as well as the skewness and kurtosis values for different sets of
parameters. It is clear that the mean, standard deviation, and upper

44
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

moments change significantly when the parameters are changed,

reflecting the distribution's flexibility in controlling the spread and
centering of the data. The skewness values exhibit both positive and
negative skewness, indicating that the distribution can represent
asymmetric data. Kurtosis ranges from larger to smaller than normal
kurtosis, indicating the distribution's ability to represent data with heavy
or light tails. Together, these results demonstrate that the NOWIBX
distribution is capable of fitting a variety of data shapes by adjusting its
parameters.
The moment generating function (𝑚𝑔𝑓) is given by the form using
equation (2,15) and equation (3.14):
𝑀𝑥 (𝑦)𝐻𝑊𝐼𝐵𝑋
𝑟
∞ 𝛼 2𝛾𝑦 𝑛 𝑟
=∑ [Θ ∑ ( ) (−1)𝑟 𝐸𝑞 (𝑔(𝑋))
𝑛=0 𝑛! 𝑖
𝑖=0 (3.17)
𝑟
𝑟
− Β ∑ ( 𝑗 ) (−1)𝑟 (𝑤 + 𝛾 + 1)𝑟−𝑗 𝐸𝑞 (𝑘 (𝑋))]
𝑗=0

3.3.4 Probability-Weighted Moments

The probability-weighted moments of NOWIBX distribution can be
found using the equation (2.17) and (3.10) to get:
∞
𝛼2 𝛼2 𝛾(ℎ+𝑠)−1
𝑝−3 −(𝑝+1)𝑥 2 −
𝜏𝑞,𝑠 = Θ𝛫2𝛼 2𝛾 ∫ 𝑥 𝑒 (1 − 𝑒 𝑥2 ) 𝑑𝑥
0

∞ 𝛾(𝑠+1)−1
𝛼2 𝛼2
𝑝−3 −(𝑤+𝛾+1+𝑝)𝑥 2 −
−Β𝛫2𝛼 2 𝛾 ∫ 𝑥 𝑒 (1 − 𝑒 𝑥2 ) 𝑑𝑥
0

In the same way as proving the moment function of the distribution,

we obtain the function in the form:

45
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝑝
𝑝
𝜏𝑝,𝑠 = Θ𝛫𝛼 2𝛾 ∑ ( ) (−1)𝑝 (𝑝 + 1)𝑝−𝑖 𝐸𝑞 (𝑔(𝑋))
𝑖
𝑖=0
𝑝
(3.18)
𝑝
−Β𝛫𝛼 2𝛾 ∑ ( 𝑗 ) (−1)𝑝 (𝑤 + 𝛾 + 𝑝 + 1)𝑝−𝑗 𝐸𝑞 (𝑘 (𝑋))
𝑗=0

3.3.5 Rényi Entropy

The Rényi entropy for the NOWIBX distribution can be found by
using the equation (2.19) and (3.11) to get:
∞ 𝛾(ℎ+𝑠)−1
1 𝛼2 𝛼2
2 −3 − 𝑥 2 − 2
𝐼𝑐 (𝑥 )𝐻𝑊𝐼𝐵𝑋 = 𝑙𝑜𝑔 ∫ (Θ2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥 )
1−𝑐
0
𝑐
𝛼2 𝛼2 𝛾(𝑠+1)−1
2 −3 −(𝑤+𝛾+1)𝑥 2 −
−Β 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥2 ) ) 𝑑𝑥

Using the expansion of the above equation as follows:

∞
𝛼2 𝛾(ℎ+𝑠)−1 𝛾 (ℎ + 𝑠) − 1 −𝑝𝛼22
−
(1 − 𝑒 𝑥2 ) = ∑ (−1 )𝑝 ( )𝑒 𝑥
𝑝
𝑝=0
𝛾(𝑠+1)−1 ∞
𝛼2 𝛾(𝑠 + 1) − 1 −𝑞𝛼22
−
(1 − 𝑒 𝑥2 ) = ∑(−1 )𝑞 ( )𝑒 𝑥
𝑞
𝑞=0
Then we get:
𝛼2 𝛼2
−3 −(𝑝+1)𝑥 2 −3 −(𝑤+𝛾+𝑞+1)𝑥 2 (3.19)
𝑓 (𝑥 ) = 𝐴𝑥 𝑒 − 𝐵𝑥 𝑒
where
𝐴=
(−1)𝑖(𝜃+1)+𝜃−1+𝑧+𝑠+ℎ+𝑝
∑∞
𝑖,𝑧,𝑠,ℎ,𝑝=0 𝛿 𝑖+1𝐷𝜃(𝑖+1)−1,𝑧 (𝑧𝑠)(ℎ1) (𝛾(ℎ+𝑠)−1
𝑝
) 𝜃2𝛼 2𝛾
𝑖!
And
𝐵=
(−1)𝑖(𝜃+1)+𝜃−1+𝑧+𝑠+𝑤+𝑞
∑∞
𝑖,𝑧,𝑠,𝑤,𝑞=0 𝛿 𝑖+1𝐷𝛾,𝑤 𝐷𝜃(𝑖+1)−1,𝑧 (𝑧𝑠) (𝛾(𝑠+1)−1
𝑞
) 𝜃2𝛼 2𝛾
𝑖!
By substituting equation (3.19) we get:

46
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

∞ 𝑐
1 𝛼2 𝛼2
−3 −(𝑝+1)𝑥 2 −3 −(𝑤+𝛾+𝑞+1)𝑥 2
𝐼𝑐 (𝑥 )𝐻𝑊𝐼𝐵𝑋 = 𝑙𝑜𝑔 ∫ (𝐴𝑥 𝑒 − 𝐵𝑥 𝑒 ) 𝑑𝑥
1−𝑐
0

By using binomial series expansion as 𝜋 = 𝑘 (𝑝 − 𝑤 − 𝛾 − 𝑞) +

𝑐 (𝑤 + 𝛾 + 𝑞 − 1), we get:
𝑐 ∞ 𝛼2
−𝜋
1 𝑐 𝑒 𝑥2
𝐼𝑐 (𝑥 )𝐻𝑊𝐼𝐵𝑋 = log [∑ (−1)𝑘 ( ) 𝐴. 𝐻 ∫ 3𝑐 𝑑𝑥]
1−𝑐 𝑘 𝑥
𝑘=0 0

3𝑐−1
1 𝛼−3𝑐−1 (−1)𝑘+1 (𝑘𝑐 )𝐴.𝐻Γ(
)
𝐼𝑐 (𝑥 )𝐻𝑊𝐼𝐵𝑋 = log [∑𝑐𝑘=0 2
3𝑐−1 ] (3.20)
1−𝑐 2[𝑘(𝑝−𝑤−𝛾−𝑞)+𝑐(𝑤+𝛾+𝑞−1)] 2

3.3.6 Order statistics

The pdf for the 𝑗-𝑡ℎ order statistics of a random sample of size 𝑛 from
the NOWIBX distribution can be expressed as:
𝑛−𝑗 𝑛−𝑗 𝜃 𝑗+𝑟−1
𝑓𝑗:𝑛 (𝑥 ) = ∑ 𝑘 (−1)𝑟 () [1 − 𝑒 −𝛿[−(1−𝜑) log 𝜑] ]
𝑟=0 𝑟
𝛼 2 𝛼2 𝛾−1 1 − 𝜑 (3.21)
2 −3 − 𝑥 2 − 2
. 𝛿𝜃 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 𝑥 ) [ − log 𝜑]
𝜑
𝜃
. [−(1 − 𝜑)𝑙𝑜𝑔 𝜑]𝜃−1𝑒 −𝛿[−(1−𝜑) log 𝜑]
𝛼2
𝛾
−
where 𝜑 = (1 − 𝑒 𝑥2 )

To obtain the order statistics of the minimum, we substitute 𝑗 = 1 into

equation (3.21) to find 𝑓𝑗:𝑛 (𝑥 ). Conversely, to obtain the order statistics
of the maximum, we substitute 𝑗 = 𝑛.

3.4 Estimation
3.4.1 Maximum Likelihood Estimation (MLE)
The parameters of the NOWIBX distribution are determined using the
maximum likelihood estimation approach. The log-likelihood function for
a random sample 𝑥1 , 𝑥2 , … , 𝑥𝑛 is obtained. The distribution adheres to the
pdf of the NOWIBX distribution [44].

47
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝐿( Φ, 𝑥 ) = ∏ 𝑓(𝑥, 𝛿, 𝜃, 𝛼, 𝛾)
𝑖=1
𝛼2 𝛾
−
𝑛 𝛾−1 1 − (1 − 𝑒 𝑥2 )
𝛼2 𝛼2
2 −3 − 𝑥 2 −
𝐿( Φ, 𝑥𝑖 ) = ∏ 𝛿𝜃 2𝛼 𝛾𝑥 𝑒 (1 − 𝑒 2
𝑥 )
𝛼2 𝛾
𝑖=1 −
(1 − 𝑒 𝑥2 )
[
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
𝛼2 𝛾 −𝛿[−(1−(1−𝑒 𝑥 ) ) log(1−𝑒 𝑥 ) ]
−
− log (1 − 𝑒 𝑥2 ) 𝑒

]
𝛾 𝛾 𝜃−1
𝛼2 𝛼2
− 2 −
. [− (1 − (1 − 𝑒 𝑥 ) ) 𝑙𝑜𝑔 (1 −𝑒 𝑥2 ) ]

The natural log-likelihood function L is obtained as:

𝐿 = 𝑛𝑙𝑜𝑔 (𝛿 ) + 𝑛𝑙𝑜𝑔 (𝜃 ) + 𝑛𝑙𝑜𝑔2 + 2𝑛𝑙𝑜𝑔 (𝛼 ) + 𝑛𝑙𝑜𝑔 (𝛾)

𝑛
𝛼2 𝑛 𝛼2
− 2
−∑ + (𝛾 − 1) ∑ log [1 − 𝑒 𝑥 ]
𝑥𝑖 𝑖=1
𝑖=1
𝛼2 𝛾
−
1 − (1 − 𝑒 𝑥2 )
𝑛 𝛼2 𝛾
−
+∑ log 𝛾 − log (1 − 𝑒 𝑥2 )
𝑖=1 𝛼2 (3.22)
−
(1 − 𝑒 𝑥2 )
[ ]
𝛾 𝛾
𝑛 𝛼2 𝛼2
− −
+(𝜃 − 1) ∑ log [− (1 − (1 − 𝑒 𝑥 2 ) ) log (1 −𝑒 𝑥2 ) ]
𝑖=1

𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿 ∑𝑛𝑖=1 [[− (1 − (1 − 𝑒 𝑥 ) ) log (1 − 𝑒 𝑥 ) ]]

To obtain the solution, we calculate the partial derivatives of the

𝜕𝐿
distribution parameters and solve the non-linear equations for =
𝜕𝛿
𝜕𝐿 𝜕𝐿 𝜕𝐿
0, = 0, = 0, = 0 provides the maximum likelihood estimation of
𝜕𝜃 𝜕𝛼 𝜕𝛾

the parameters 𝛿, 𝜃, 𝛼, and γ, in that order. The only way to solve it was

48
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

by using numerical methods. These methods utilize software like R,

MAPLE, SAS, and other similar programs.

4.2 Ordinary Least Squares Estimation (OLSE)

The OLSE can be found by using [47]:
1 2
𝜑(𝛿, 𝜃, 𝛼, 𝛾) = ∑𝑛𝑖=1 [𝐹 (𝑥𝑖 ) − ] (3.23)
𝑛+1

𝛾 𝛾 𝜃
2
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
−𝛿 −(1−(1−𝑒 ) ) log(1−𝑒 )
1
𝜑(∅) = ∑𝑛𝑖=1 1 − 𝑒 [ ] − (3.24)
𝑛+1

[ ]
where ∅ = (𝛿, 𝜃, 𝛼, 𝛾)
By partially deriving the above equation for the 𝛿, 𝜃, 𝛼, 𝛾 parameters,
we get:
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
𝑛 −𝛿 −(1−(1−𝑒 ) ) log(1−𝑒 )
𝜕𝜑 1
= 2∑ 1 −𝑒 [ ] −
𝜕𝛿 𝑛+1
𝑖=1

[ ]
𝛼2
𝛾
𝛼2
𝛾 𝜃 (3.25)
− 2 − 2
𝑥𝑖 𝑥𝑖
−𝛿 − 1− 1−𝑒 log 1−𝑒

𝜕 −𝑒 [ ( ( ) ) ( ) ]

( )
𝜕𝛿
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
𝑛 −𝛿 −(1−(1−𝑒 ) ) log(1−𝑒 )
𝜕𝜑 1
= 2∑ 1 −𝑒 [ ] − (3.26)
𝜕𝜃 𝑛+1
𝑖=1

[ ]

49
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
−𝛿 − 1− 1−𝑒 log 1−𝑒

𝜕 −𝑒 [ ( ( ) ) ( ) ]

( )
𝜕𝜃
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
𝑛 −𝛿 −(1−(1−𝑒 ) ) log(1−𝑒 )
𝜕𝜑 1
= 2∑ 1 −𝑒 [ ] −
𝜕𝛼 𝑛+1
𝑖=1

[ ]
𝛼2
𝛾
𝛼2
𝛾 𝜃 (3.27)
− 2 − 2
𝑥𝑖 𝑥𝑖
−𝛿 − 1− 1−𝑒 log 1−𝑒

𝜕 −𝑒 [ ( ( ) ) ( ) ]

( )
𝜕𝛼
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
𝑛 −𝛿 −(1−(1−𝑒 ) ) log(1−𝑒 )
𝜕𝜑 1
= 2∑ 1 −𝑒 [ ] −
𝜕𝛾 𝑛+1
𝑖=1

[ ]
𝛾 𝛾 𝜃 (3.28)
𝛼2 𝛼2
− 2 − 2
𝑥𝑖 𝑥𝑖
−𝛿 − 1− 1−𝑒 log 1−𝑒

𝜕 −𝑒 [ ( ( ) ) ( ) ]

( )
𝜕𝛾

By setting the previous equations equal to zero, we obtain the OLSE:

𝜕𝐿 𝜕𝐿 𝜕𝐿 𝜕𝐿
= 0, = 0, = 0, =0
𝜕𝛿 𝜕𝜃 𝜕𝛼 𝜕𝛾
We note that the equations are equal to zero. It is clear that it is not
possible to obtain the closed form of the above equations and it is
difficult to solve them manually. Therefore, it is necessary to use

50
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

computer programs or numerical methods to find an estimate of these

parameters.

4.3 Weighted Least Squares Estimators (WLSE)

The weighted least squares estimators can be obtained by the equation
[47]:
(𝑛+1)2 (𝑛+2) 𝑖 2
𝜔 (𝛿, 𝜃, 𝛼, 𝛾) = ∑𝑛𝑖=1 [𝐹 (𝑥𝑖 ) − ] (3.29)
𝑖(𝑛−𝑖+1) 𝑛+1

𝑛
(𝑛 + 1)2 (𝑛 + 2)
𝜔 (∅) = ∑ 1
𝑖(𝑛 − 𝑖 + 1)
𝑖=1

[
2 (3.30)
𝛾 𝛾 𝜃
𝛼2 𝛼2
− 2 − 2
−𝛿 −(1−(1−𝑒 𝑥𝑖 ) ) log(1−𝑒 𝑥𝑖 )
𝑖
−𝑒 [ ] −
𝑛+1

]
In the same way as MLE and OLSE, the above equation is derived for
the NOWIBX distribution parameters and equalized to zero to obtain
parameter estimates using the WLSE method.

3.5 Simulation Study

The performance of MLE, OLSE, WLSE, ADE, and RTADE for the
NOWIBX distribution is evaluated through a Monte Carlo simulation
study using the R software. The sample sizes considered in the study are
n = 30,60,120, 180, 240, and 300. We generate N = 1000 samples for the
true parameter values listed in Tables (3.3) and (3.4). The resulting MLEs
for the model parameters are averaged to obtain the mean values. The
corresponding bias and RMSE are calculated as well. The bias and RMSE
for a specific estimated parameter, denoted as 𝜆̂, are given by [40]:

51
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

∑ 𝑁 ̂
λ
𝑏𝑖𝑎𝑠(λ̂) = 𝑖=1 𝑖 − λ (3.31)
𝑁

2
∑𝑁 ̂
̂ √ 𝑖=1(λ𝑖 − λ) (3.32)
𝑅𝑀𝑆𝐸(λ) =
𝑁
The results indicate the consistency of all estimators presented in
Tables. (3.3) and (3,4). As the sample size increases, the average
parameter estimations converge towards the true parameter values.
Furthermore, the MSEs decrease in magnitude as the sample size grows.
Table (3.3) : Monte Carlo simulations-1 conducted for the NOWIBX
𝜹 =𝟏 , 𝜽 =𝟏. 𝟑, 𝜶 =𝟏. 𝟕, 𝜸 = 𝟏. 𝟒
N Est. Ess. Par. MLE OLSE WLSE ADE RTADE
𝛿̂ 0.84851 1.29930 1.23168 0.94696 1.29166
𝜃̂ 1.24517 1.01519 1.08886 1.13436 0.94703
Mean
𝛼̂ 2.25033 1.91147 1.94887 2.31510 1.92138
𝛾̂ 1.38846 1.35122 1.33243 1.37331 1.37025
𝛿̂ 0.20155 0.80347 0.58634 0.77941 0.74152
𝜃̂ 2.58999 0.58985 0.87494 1.78335 0.55112
MSE
𝛼̂ 0.93015 0.40443 0.40086 1.09267 0.39930
𝛾̂ 0.20769 0.16801 0.12084 0.23189 0.16411
30
𝛿̂ 0.44894 0.89636 0.76573 0.88284 0.86111
𝜃̂ 1.60934 0.76802 0.93538 1.33542 0.74237
RMSE
𝛼̂ 0.96444 0.63595 0.63313 1.04530 0.63190
𝛾̂ 0.45573 0.40990 0.34762 0.48155 0.40511
𝛿̂ 0.15148 0.29930 0.23168 0.05303 0.29166
𝜃̂ 0.05482 0.28480 0.21113 0.16563 0.35296
Bias
𝛼̂ 0.55033 0.21147 0.24887 0.61510 0.22138
𝛾̂ 0.01153 0.04877 0.06756 0.02668 0.02974
𝛿̂ 0.96710 1.10466 1.06691 1.19665 1.04953
𝜃̂ 1.49976 1.01804 1.17710 1.20604 1.17942
Mean
𝛼̂ 2.03668 1.97248 1.93637 1.87302 1.89220
𝛾̂ 1.39320 1.30122 1.33059 1.33028 1.382113
𝛿̂ 0.19043 0.21288 0.19437 0.30065 0.12793
𝜃̂ 2.32695 0.48608 0.69203 0.60688 0.62783
MSE
60 𝛼̂ 0.75296 0.28997 0.29208 0.29954 0.23421
𝛾̂ 0.16475 0.07159 0.08587 0.08191 0.10748
𝛿̂ 0.43638 0.46139 0.44088 0.54831 0.35768
𝜃̂ 1.52543 0.69719 0.83188 0.77902 0.79235
RMSE
𝛼̂ 0.86773 0.53849 0.54044 0.54730 0.48395
𝛾̂ 0.40589 0.26756 0.29304 0.28620 0.32785
Bias 𝛿̂ 0.03289 0.10466 0.06691 0.19665 0.04953

52
CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝜃̂ 0.19976 0.28195 0.12289 0.09395 0.12057

𝛼̂ 0.33668 0.27248 0.23637 0.17302 0.19220
𝛾̂ 0.00679 0.09877 0.06940 0.06971 0.01788
𝛿̂ 0.96317 0.99527 1.05478 1.09483 1.10011
𝜃̂ 1.50624 1.05935 1.31618 1.06764 1.27253
Mean
𝛼̂ 1.93978 1.97610 1.85190 1.91211 1.87369
𝛾̂ 1.42259 1.31919 1.35521 1.34185 1.31936
𝛿̂ 0.10619 0.06944 0.09112 0.16966 0.13872
𝜃̂ 2.28356 0.41757 0.63192 0.53797 0.58790
MSE
𝛼̂ 0.52845 0.21440 0.23865 0.20876 0.19866
𝛾̂ 0.14471 0.04831 0.09763 0.07770 0.04800
120
𝛿̂ 0.32587 0.26352 0.30186 0.41190 0.37246
𝜃̂ 1.51114 0.64619 0.79493 0.73346 0.76675
RMSE
𝛼̂ 0.72694 0.46303 0.48852 0.45690 0.44571
𝛾̂ 0.38041 0.21980 0.31246 0.27875 0.21910
𝛿̂ 0.03682 0.00472 0.05478 0.09483 0.10011
𝜃̂ 0.20624 0.24064 0.01618 0.23235 0.02746
Bias
𝛼̂ 0.23978 0.27610 0.15190 0.21211 0.17369
𝛾̂ 0.02259 0.08080 0.04478 0.05814 0.08063
𝛿̂ 1.00079 1.00184 0.97161 0.95509 0.97417
𝜃̂ 1.63097 1.11325 1.19447 1.12252 1.39367
Mean
𝛼̂ 1.84001 1.88854 1.93437 1.85846 1.85203
𝛾̂ 1.41865 1.37515 1.34733 1.38554 1.33427
𝛿̂ 0.09266 0.05298 0.05155 0.05570 0.07632
𝜃̂ 1.99584 0.40910 0.49488 0.23102 0.64039
MSE
𝛼̂ 0.45411 0.19982 0.24237 0.12586 0.18974
𝛾̂ 0.11620 0.06271 0.04969 0.03858 0.03986
180
𝛿̂ 0.30441 0.23017 0.22704 0.23602 0.27626
𝜃̂ 1.41274 0.63961 0.70347 0.48064 0.80024
RMSE
𝛼̂ 0.67388 0.44701 0.49231 0.35476 0.43560
𝛾̂ 0.34089 0.25043 0.22291 0.19641 0.19965
𝛿̂ 0.00079 0.00184 0.02838 0.04490 0.02582
𝜃̂ 0.33097 0.18674 0.10552 0.17747 0.09367
Bias
𝛼̂ 0.14001 0.18854 0.23437 0.15846 0.15203
𝛾̂ 0.01865 0.02484 0.05266 0.01445 0.06572
𝛿̂ 1.01194 0.99087 0.99696 0.99555 0.98553
𝜃̂ 1.60314 1.10873 1.19984 1.10360 1.31456
Mean
𝛼̂ 1.83614 1.94423 1.89376 1.93047 1.86391
𝛾̂ 1.41506 1.31951 1.34727 1.31984 1.33997
240 𝛿̂ 0.08895 0.04816 0.04870 0.05875 0.03883
𝜃̂ 1.83304 0.28415 0.49931 0.34089 0.37508
MSE
𝛼̂ 0.38765 0.19969 0.18497 0.18138 0.19250
𝛾̂ 0.12716 0.03682 0.04589 0.05106 0.03930
RMSE 𝛿̂ 0.29825 0.21945 0.22069 0.24238 0.19705

53
CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝜃̂ 1.35389 0.53305 0.70661 0.58385 0.61244

𝛼̂ 0.62262 0.44687 0.43008 0.42589 0.43875
𝛾̂ 0.35660 0.19190 0.21422 0.22597 0.198262
𝛿̂ 0.01194 0.00912 0.00303 0.00444 0.01446
𝜃̂ 0.30314 0.19126 0.10015 0.19639 0.01456
Bias
𝛼̂ 0.13614 0.24423 0.19376 0.23047 0.16391
𝛾̂ 0.01506 0.08048 0.05272 0.08015 0.06002
𝛿̂ 1.01905 1.00140 0.99236 1.02010 1.03990
𝜃̂ 1.63442 1.17367 1.29927 1.20270 1.39133
Mean
𝛼̂ 1.81403 1.86004 1.83932 1.91986 1.72402
𝛾̂ 1.42553 1.36383 1.35078 1.28994 1.41140
𝛿̂ 0.09253 0.04250 0.03445 0.03007 0.04057
𝜃̂ 2.00065 0.53160 0.45079 0.38939 0.43030
MSE
𝛼̂ 0.36295 0.12173 0.13315 0.17983 0.11702
𝛾̂ 0.14291 0.03691 0.03233 0.03530 0.04695
300
𝛿̂ 0.30419 0.20617 0.18562 0.17340 0.20142
𝜃̂ 1.41444 0.72911 0.67141 0.62401 0.655976
RMSE
𝛼̂ 0.60245 0.34890 0.36491 0.42406 0.34209
𝛾̂ 0.37804 0.19212 0.17980 0.18788 0.21668
𝛿̂ 0.01905 0.00140 0.00763 0.02010 0.03990
𝜃̂ 0.33442 0.12632 0.00072 0.09729 0.09133
Bias
𝛼̂ 0.11403 0.16004 0.13932 0.21986 0.02402
𝛾̂ 0.02553 0.03616 0.04921 0.11005 0.01140

Table (3.4) : Monte Carlo simulations-2 conducted for the NOWIBX

𝜹 =𝟏. 𝟏 , 𝜽 =𝟏. 𝟓, 𝜶 =𝟏. 𝟕, 𝜸 = 𝟎. 𝟖
N Est. Ess. Par. MLE OLSE WLSE ADE RTADE
𝛿̂ 0.93133 1.30795 1.38786 0.83780 1.37114
𝜃̂ 1.39481 1.14878 1.11957 1.06747 1.12095
Mean
𝛼̂ 2.28297 1.95588 1.90918 2.58867 1.92388
𝛾̂ 0.85159 0.77909 0.82584 0.80526 0.82210
𝛿̂ 0.30417 0.63838 0.93775 0.34260 0.92326
𝜃̂ 7.36303 0.71589 0.68857 2.09830 0.65250
MSE
𝛼̂ 1.18834 0.39762 0.43739 1.87058 0.46158
𝛾̂ 0.12910 0.04264 0.08567 0.10167 0.07645
30
𝛿̂ 0.55151 0.79899 0.96837 0.58532 0.96086
𝜃̂ 2.71349 0.84610 0.82980 1.44855 0.80777
RMSE
𝛼̂ 1.09011 0.63057 0.66135 1.36769 0.67939
𝛾̂ 0.35931 0.20650 0.29271 0.31887 0.27650
𝛿̂ 0.16866 0.20795 0.28786 0.26219 0.27114
𝜃̂ 0.10518 0.35121 0.38042 0.43252 0.37905
Bias
𝛼̂ 0.58297 0.25588 0.20918 0.88867 0.22388
𝛾̂ 0.05159 0.02090 0.02584 0.00526 0.02210
60 Mean 𝛿̂ 1.38684 1.68647 1.75331 1.82507 0.38186

54
CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝜃̂ 0.97228 1.27315 1.39812 1.29063 0.60213

𝛼̂ 1.80620 1.72682 1.74753 1.77267 1.14698
𝛾̂ 1.09209 0.82331 0.84125 0.86543 0.18196
𝛿̂ 1.03405 2.09718 2.54195 2.67515 0.53513
𝜃̂ 1.05040 0.76080 1.04376 0.87526 0.86511
MSE
𝛼̂ 0.09266 0.16299 0.15628 0.15207 1.16036
𝛾̂ 0.20429 0.06610 0.12941 0.08250 0.41636
𝛿̂ 1.01688 1.44816 1.59435 1.63559 0.73152
𝜃̂ 1.02489 0.87224 1.02164 0.93555 0.93011
RMSE
𝛼̂ 0.30440 0.40372 0.39532 0.38997 1.07720
𝛾̂ 0.45198 0.25711 0.35974 0.28723 0.64526
𝛿̂ 0.28684 0.58647 0.65331 0.72507 0.71813
𝜃̂ 0.52771 0.22684 0.10187 0.20936 0.89786
Bias
𝛼̂ 0.10620 0.02682 0.04753 0.07267 0.55301
𝛾̂ 0.29209 0.02331 0.04125 0.06543 0.61803
𝛿̂ 1.19949 1.50905 1.40785 1.46379 0.35964
𝜃̂ 1.10817 1.40081 1.45263 1.42224 0.64666
Mean
𝛼̂ 1.76531 1.72772 1.74643 1.76849 1.08244
𝛾̂ 1.03685 0.82951 0.85754 0.88262 0.16978
𝛿̂ 0.50778 1.81559 0.70640 0.99215 0.55545
𝜃̂ 0.83433 0.66542 0.84138 0.69431 0.77131
MSE
𝛼̂ 0.04884 0.09978 0.08830 0.08870 0.61152
𝛾̂ 0.14617 0.05192 0.06661 0.10166 0.41163
120
𝛿̂ 0.71258 1.34744 0.84047 0.99606 0.74529
𝜃̂ 0.91342 0.81573 0.91727 0.83325 0.87824
RMSE
𝛼̂ 0.22100 0.31588 0.29716 0.29783 0.78200
𝛾̂ 0.38232 0.22787 0.25809 0.31884 0.64158
𝛿̂ 0.09949 0.40905 0.30785 0.36379 0.74035
𝜃̂ 0.39182 0.09918 0.04736 0.07775 0.85333
Bias
𝛼̂ 0.06531 0.02772 0.04643 0.06849 0.61755
𝛾̂ 0.23685 0.02951 0.05754 0.08262 0.63021
𝛿̂ 1.19260 1.50469 1.29586 1.40195 0.34662
𝜃̂ 1.22710 1.30195 1.42239 1.33533 0.64321
Mean
𝛼̂ 1.74438 1.78584 1.72547 1.76609 1.08354
𝛾̂ 0.97447 0.87144 0.86443 0.87352 0.16241
𝛿̂ 0.45424 0.82283 0.54927 0.76344 0.57323
𝜃̂ 0.82713 0.53538 0.67489 0.55264 0.75792
MSE
180 𝛼̂ 0.03206 0.07341 0.06013 0.05450 2.53318
𝛾̂ 0.10542 0.05973 0.06988 0.04133 0.44806
𝛿̂ 0.67397 0.90710 0.74113 0.87375 0.75712
𝜃̂ 0.90946 0.73170 0.82151 0.74340 0.87058
RMSE
𝛼̂ 0.17907 0.27095 0.24523 0.23346 1.59159
𝛾̂ 0.32468 0.24441 0.26434 0.20331 0.66937
Bias 𝛿̂ 0.09260 0.40469 0.19586 0.30195 0.75337

55
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

𝜃̂ 0.27289 0.19804 0.07760 0.16466 0.85678

𝛼̂ 0.04438 0.08584 0.02547 0.06609 0.61645
𝛾̂ 0.17447 0.07144 0.06443 0.07352 0.63758
𝛿̂ 1.16899 1.32617 1.30655 1.25123 0.34158
𝜃̂ 1.31102 1.35640 1.46087 1.41861 0.65988
Mean
𝛼̂ 1.73442 1.74811 1.74580 1.71891 1.01333
𝛾̂ 0.93586 0.87285 0.85538 0.85568 0.15154
𝛿̂ 0.29801 0.50504 0.37666 0.62630 0.57926
𝜃̂ 0.64330 0.48666 0.67974 0.42511 0.72772
MSE
𝛼̂ 0.02637 0.06027 0.05227 0.05164 0.50527
𝛾̂ 0.08694 0.05335 0.04908 0.03456 0.42258
240
𝛿̂ 0.54590 0.71066 0.61373 0.79139 0.76109
𝜃̂ 0.80206 0.69761 0.82446 0.65200 0.85306
RMSE
𝛼̂ 0.16240 0.24550 0.22863 0.22724 0.71082
𝛾̂ 0.29487 0.23097 0.22156 0.18590 0.65006
𝛿̂ 0.06899 0.22617 0.20655 0.15123 0.75841
𝜃̂ 0.18897 0.14359 0.03912 0.08138 0.84011
Bias
𝛼̂ 0.03442 0.04811 0.04580 0.01891 0.68666
𝛾̂ 0.13586 0.07285 0.05538 0.05568 0.64845
𝛿̂ 1.16432 1.20899 1.29348 1.20446 0.33885
𝜃̂ 1.37156 1.41505 1.46209 1.42598 0.64934
Mean
𝛼̂ 1.72514 1.71557 1.73806 1.69934 1.00647
𝛾̂ 0.90857 0.85282 0.84609 0.83187 0.14769
𝛿̂ 0.28228 0.28998 0.25128 0.33052 0.58261
𝜃̂ 0.54583 0.39820 0.62764 0.35156 0.73950
MSE
𝛼̂ 0.02139 0.05498 0.04632 0.03659 0.50218
𝛾̂ 0.07777 0.04380 0.04823 0.03199 0.42640
300
𝛿̂ 0.53130 0.53850 0.50128 0.57491 0.76329
𝜃̂ 0.73880 0.63103 0.79224 0.59293 0.85994
RMSE
𝛼̂ 0.14626 0.23449 0.21523 0.19130 0.70864
𝛾̂ 0.27888 0.20930 0.21961 0.17886 0.65299
𝛿̂ 0.06432 0.10899 0.19348 0.10446 0.76114
𝜃̂ 0.12843 0.08494 0.03790 0.07401 0.85065
Bias
𝛼̂ 0.02514 0.01557 0.03806 0.00065 0.69352
𝛾̂ 0.10857 0.05282 0.04609 0.03187 0.65230

Table 3.3 displays the results of the first Monte Carlo simulation of
the NOWIBX distribution at the true values of the parameters 𝛿 = 1, 𝜃 =
1.3, 𝛼 = 1.7, and 𝛾 = 1.4 using five estimation methods: MLE, OLSE,
WLSE, ADE, and RTADE, and for different levels of sample size (N).
The outputs include the estimated parameter means, MSE, RMSE, and

56
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

Bias. It can be seen that increasing the sample size generally reduces the
MSE and RMSE values for all methods, reflecting improved estimation
accuracy with more data. The bias also approaches zero at large 𝑁,
indicating the consistency of these estimates. It appears that some
methods, such as WLSE and OLSE, achieve lower MSE than MLE in
many cases, especially at small and medium 𝑁, demonstrating their
relative efficiency in these conditions. The estimated mean values
gradually approach the true values as 𝑁 increases, a behavior expected in
consistent estimations.
Table 3.4 presents the same type of analysis but for different true
values of the parameters 𝛿 = 1.1, 𝜃 = 1.5, 𝛼 = 1.7, and 𝛾 = 0.8. The
results confirm the general trend, with MSE and RMSE decreasing as N
increases, and convergence to the true values gradually improves.
However, we note that some parameters, particularly δ and θ, exhibit
larger differences between methods at small sample sizes. MLE may
exhibit greater bias than OLSE or WLSE, while the latter maintains
relative stability. Furthermore, there are instances at small N values
where differences in MSE and RMSE between methods are evident,
indicating that the choice of estimation method becomes important at this
stage. The general trend in both tables reflects that the efficiency of the
methods improves with increasing sample size, and that WLSE and
OLSE often perform similarly to, or even better than, MLE at small
sample sizes.

3.6 Application
In order to show the efficacy of the NOWIBX distribution in
accurately fitting data, we provide a real-world example using two data
sets. The objective is to highlight the benefits of NOWIBX and the

57
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

degree to which it aligns with the data. Table (3.5) presents a comparative
analysis of NOWIBX and various distributions for the utilized data.

Table (3.5). CDF functions for comparative distributions

Distribution CDF

𝛾 𝜃
𝛼2
[0,1] Truncated Nadarajah-Haghighi Inverse 1−
−
1+𝛿(1−(1−𝑒 𝑥2 ) )
Burr X ([0,1]NHIBX)
1−𝑒 ( )
𝜃
1 − 𝑒1−(1−𝛿)
θ
Kumaraswamy Inverse Burr X (KuIBX) 𝛾 δ
𝛼2
−
(New) 1 − (1 − (1 − (1 − 𝑒 𝑥2 ) ) )

θ
Exponential Generalized Inverse Burr X 𝛾 δ
𝛼2
− 2
(EGIBX) (New) (1 − ( 1 − (1 − 𝑒 𝑥 ) ) )

𝛾
Log Gamma Inverse Burr X (LGamIBX) 𝛼2
− 2
1 − Γ (−𝛿 log ((1 −𝑒 𝑥 ) ) , 𝜃)

𝛾
Beta Inverse Burr X (BeIBX) 𝛼2
2 − 2
𝑝𝛼 𝛾(1 − (1 −𝑒 𝑥 ) , δ, θ)

𝛾
𝛼2
Inverse Burr X (IBX) − 2
1 − (1 − 𝑒 𝑥 )

This comparison is conducted using four goodness-of-fit statistics are

𝐾𝑆, A, W, and p-value and four information criteria HQIC [45], BIC,
AIC, and CAIC [46]. These measures are commonly employed to assess
the quality of fit.
Table (3.6) and Table (3.7) display the NOWIBX distribution which
has the lowest values of AIC, AICC, and BIC when compared to the
corresponding values of non-overlapping distributions. In addition, the
goodness-of-fit statistics A, W, 𝐾𝑆 tests, and p-value all suggest that the
NOWIBX distribution is the most suitable fit for both data I and data II.
 The First Dataset I
This application focuses on the duration of remission (in months) for a
group of 128 patients who have been diagnosed with bladder cancer. This

58
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

data has undergone recent analysis in numerous studies [47]. The dataset
contains the following values:
0.08 2.09 3.48 4.87 6.94 8.66 13.11 23.63 0.20 2.23
3.52 4.98 6.97 9.02 13.29 0.40 2.26 3.57 5.06 7.09
9.22 13.80 25.74 0.50 2.46 3.64 5.09 7.26 9.47 14.24
25.82 0.51 2.54 3.70 5.17 7.28 9.74 14.76 26.31 0.81
2.62 3.82 5.32 7.32 10.06 14.77 32.15 2.64 3.88 5.32
7.39 10.34 14.83 34.26 0.90 2.69 4.18 5.34 7.59 10.66
15.96 36.66 1.05 2.69 4.23 5.41 7.62 10.75 16.62 43.01
1.19 2.75 4.26 5.41 7.63 17.12 46.12 1.26 2.83 4.33
5.49 7.66 11.25 17.14 79.05 1.35 2.87 5.62 7.87 11.64
17.36 1.40 3.02 4.34 5.71 7.93 11.79 18.10 1.46 4.40
5.85 8.26 11.98 19.13 1.76 3.25 4.50 6.25 8.37 12.02
2.02 3.31 4.51 6.54 8.53 12.03 20.28 2.02 3.36 6.76
12.07 21.73 2.07 3.36 6.93 8.65 12.63 22.69

n mean sd median trimmed mad min max range skew KU SE

128 9.37 10.51 6.39 7.42 5.46 0.08 79.05 78.97 3.25 15.2 0.93

Table (3.6): Information criteria values for comparative distributions for data I
Dist. -L AIC CAIC BIC HQIC
NOWIBX 412.1536 832.3073 832.6325 843.7154 836.9424
[0,1]NHIBX 461.7729 931.5616 931.8868 942.9697 936.1968
KuIBX 415.2403 838.4805 838.8057 849.8886 843.1157
EGIBX 447.3125 902.625 902.9502 914.0331 907.2602
LGamIBX 425.1468 858.2937 858.6189 869.7018 862.9288
BeIBX 424.5922 857.1844 857.5096 868.5925 861.8196
IBX 505.1199 1014.24 1014.336 1019.944 1016.557

Table (3.7): Evaluate goodness-of-fit statistics for comparative distributions for data I
Dist. W A 𝑲𝑺 p-value
NOWIBX 0.06721345 0.4552835 0.05229047 0.8752071
[0,1]NHIBX 0.7652793 4.72932 0.3145878 1.986578e-11
KuIBX 0.1180385 0.8011623 0.06114451 0.724887
EGIBX 0.8020442 4.875024 0.1513152 0.005694233
LGamIBX 38.55569 250.3297 0.9999987 0
BeIBX 0.3241705 2.080104 0.1003474 0.1518128
IBX 2.12294 12.00713 0.3543045 2.209344e-14

59
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

Table (3.8): parameter estimators by MLE for comparative distributions for data I
Dist. 𝜹 𝜽 𝜶 𝜸
NOWIBX 3.260230462 4.601609001 0.008892513 0.082594196
[0,1]NHIBX 0.08993546 0.99653341 0.29647638 0.61361551
KuIBX 22.545247752 26.350177927 0.005094885 0.135104612
EGIBX 0.16824563 33.86453268 0.02153313 2.11511660
LGamIBX 27.358862809 3.095337355 0.008749932 0.680536116
BeIBX 37.645205597 18.100188640 0.005253935 0.081712914
IBX - - 0.3501005 0.1769828

Results from Table (3.6) show that, in terms of the fit criteria (AIC,
BIC, HQIC, CAIC), the NOWIBX distribution recorded the lowest values
compared to the other distributions, reflecting its clear superiority in
representing the data. It was followed in terms of performance by the
BeIBX and LGamIBX distributions, but by a significant margin, while
the IBX distribution recorded the highest values, indicating its poor fit.
Results from Table (3.7): The 𝐾𝑆 test gave the NOWIBX distribution
a very small value (0.052) with a high p-value (0.875), indicating no
significant differences between the theoretical distribution and the data, a
strong indicator of the goodness of fit. In contrast, some distributions,
such as [0,1]NHIBX and EGIBX, showed very low p-values, indicating
their failure to fit. Distributions such as KuIBX and BeIBX, on the other
hand, produced acceptable results but were less accurate than NOWIBX.
The results from Table (3.8) show that the NOWIBX distribution is
estimated to have 𝛿 = 3.26, 𝜃 = 4.60, 𝛼 ≈ 0.0089, and 𝛾 ≈ 0.0826.
These values reflect a high degree of shape flexibility, allowing it to
represent a skewed, heavy-tailed distribution. The remaining distributions
exhibited significant differences in parameter estimates, especially
KuIBX and LGamIBX, which have very large shape values, which may
limit their practical flexibility in this case.

60
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

Figure (3.5) Fitted densities for Data I

Figure (3.6) Fitted empirical CDF for Data I

In Figure (3.5), the NOWIBX curve appears to follow the shape of the
data represented by the gray histogram more accurately than the other
models, especially in the first part where most of the values are
concentrated. The remaining models either underestimate the peak or
overestimate the distribution in the tail. In Figure (3.6), the NOWIBX
curve roughly matches the empirical cumulative function (dashed black
line), indicating that the distribution captures the behavior of the data
across all ranges. The other distributions deviate from the empirical
model, especially at large values.

61
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

The numerical and graphical results confirm that the NOWIBX

distribution is the most appropriate for representing the remission
duration data in this sample, as it achieved a clear advantage in
information criteria and goodness-of-fit measures, and showed an
accurate graphical representation of both density and cumulative
probability.

 The Second Dataset II

The dataset included in this study consisted of the survival periods,
measured in years, of a cohort of 46 patients who received chemotherapy
treatment exclusively. The survival times (in years) in this data set were
previously reported for easy access [47]:
0.047 0.115 0.121 0.132 0.164 0.197 0.203 0.260 0.282 0.296
0.334 0.395 0.458 0.466 0.501 0.507 0.529 0.534 0.540 0.641
0.644 0.696 0.841 0.863 1.099 1.219 1.271 1.326 1.447 1.485
1.553 1.581 1.589 2.178 2.343 2.416 2.444 2.825 2.830 3.578
3.658 3.743 3.978 4.003 4.033 4.127

n mean sd median trimmed mad min max range skew KU SE

45 1.34 1.25 0.84 1.19 0.95 0.05 4.03 3.99 0.94 -0.45 0.19

Table (3.9): Information criteria values for comparative distributions for data II
Dist. -L AIC CAIC BIC HQIC
NOWIBX 58.2242 124.4484 125.4484 131.6751 127.1424
[0,1]NHIBX 64.62518 137.2504 138.2504 144.477 139.9444
KuIBX 59.77862 127.5572 128.5572 134.7839 130.2513
EGIBX 84.1579 176.3158 177.3158 183.5424 179.0098
LGamIBX 60.84979 129.6996 130.6996 136.9262 132.3936
BeIBX 61.83997 131.6799 132.6799 138.9066 134.374
IBX 72.14705 148.2941 148.5798 151.9074 149.6411

62
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

Table (3.10): Evaluate goodness-of-fit statistics for comparative distributions for data
II
Dist. W A 𝑲𝑺 p-value
NOWIBX 0.05396241 0.3980271 0.08529341 0.871218
[0,1]NHIBX 0.1329864 0.9634076 0.1924543 0.06197589
KuIBX 0.07235006 0.5343113 0.1026565 0.691647
EGIBX 0.5826623 3.393419 0.3264961 8.996611e-05
LGamIBX 15.18441 89.70491 0.9985948 0
BeIBX 0.1056582 0.7546332 0.1126633 0.5784162
IBX 0.3059827 1.941572 0.2253925 0.01722965

Table (3.11): parameter estimators by MLE for comparative distributions for data II
Dist. 𝜹 𝜽 𝜶 𝜸
NOWIBX 7.3109037 2.9884617 0.0074122 0.0864386
[0,1]NHIBX 14.554068575 2.602689186 0.120772617 0.004792009
KuIBX 9.474044085 9.572313591 0.008266733 0.155861632
EGIBX 2.09102739 0.25800972 0.27219926 0.05589438
LGamIBX 15.817747271 1.109489996 0.006555005 1.485922500
BeIBX 10.97833848 6.27498851 0.01318763 0.12937126
IBX - - 0.1344021 0.2715192

Results from Table (3.9) show that the NOWIBX distribution

achieved the lowest values for AIC, BIC, HQIC, and CAIC, making it the
most suitable for representing the data. KuIBX and LGamIBX followed
close behind, while EGIBX and IBX showed high values indicating poor
fit.
Results from Table (3.10) show that the NOWIBX distribution
showed values of W = 0.0539, A = 0.398, and KS = 0.085, with a high p-
value (0.871), indicating a strong fit with the data and the absence of
significant differences. Distributions such as KuIBX and BeIBX also
produced acceptable results but were less accurate, while distributions
such as EGIBX and LGamIBX produced very low p-values, indicating
that their fit was rejected.
The results from Table 3.11 show that the NOWIBX distribution
was estimated to have 𝛿 ≈ 7.31, 𝜃 ≈ 2.99, 𝛼 ≈ 0.0074, and 𝛾 ≈ 0.086,
values that demonstrate good shape flexibility and the ability to represent
a skewed distribution with moderate tails. The remaining distributions

63
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

exhibited marked differences in parameters, and some, such as

LGamIBX, had high shape values that may limit their flexibility.

Figure (3.7) Fitted densities for Data II

Figure (3.8) Fitted empirical CDF for Data II

In Figure 3.7, the NOWIBX curve follows the gray histogram

accurately, especially in the first part where the values are concentrated,
with a good representation of the decrease in density toward the tail.
Some models, such as EGIBX, overestimate the first peak and do not
follow the tail accurately.

64
 CHAPTER THREE A NEW ODD WEIBULL INVERSE BURR TYPE X, PROPERTIES WITH APPLICATION

In Figure 3.8, the NOWIBX curve is nearly identical to the

empirical cumulative function, confirming the accuracy of the fit across
the entire range of values. However, some other models deviate from the
empirical one, especially in the upper half of the values.
The results confirm that NOWIBX is the best fit for these data,
combining the best information indices, the highest p-values in the fit
tests, and a well-designed graphical representation of both the density and
cumulative function.

65
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Chapter Four

Neutrosophic New Odd Weibull inverse Rayleigh

Distribution

4.1 Introduction
This chapter proposes a continuous probability distribution called the
New Odd Weibull Inverse Rayleigh distribution (NOWIR) with three
parameters. Then this distribution is expanded to deal with Neutrosophic
logic. Suppose the Neutrosophic random variable and neutrosophic
parameters are taken based on the direct neutrosophic method, which
consists of integrating the inverse Rayleigh distribution with the New
Odd Weibull family NOWG to create the New Odd Weibull Inverse
Rayleigh distribution (NOWIR) and introduce it into neutrosophic logic.
Relying on the extended (direct) neutrosophic distribution method to
generate the neutrosophic distribution, which is called NeNOWIR for
short, each parameter of the NOWIR distribution was modified to include
the elements of certainty T, uncertainty I, and failure F, to obtain a more
flexible distribution in modeling the original data, the transformation of
this distribution is divided into neutrosophic vectors and a data set is
created using the simulation method. Some statistical and mathematical
properties of the new distribution according to neutrosophic logic, are
also presented in addition to estimating the parameters using the
maximum potential method. Finally, regarding the practical aspect, the
NeNOWIR distribution was applied to two types of neutrosophic real
data: the first was the 30-day COVID-19 data for the Netherlands, and the
second type of data was represented by interval estimates mortality rates
of infants and children under five years of age and compared the results
with six other distributions using some precision criteria.

66
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

4.2 NOWIR distribution

The Inverse Rayleigh distribution, which is defined by a single
parameter 𝑐, can be represented by the cumulative distribution function
(CDF) and probability density function (PDF). These functions can be
described using the following formulas [48]:
𝑐
− (4.1)
𝐺 (𝑥; 𝑐 ) = 𝑒 𝑥2 ,𝑥 ≥ 0 ,𝑐 > 0
2𝑐 − 𝑐2 (4.2)
𝑔(𝑥; 𝑐 ) = 𝑒 𝑥 ,𝑥 ≥ 0 ,𝑐 > 0
𝑥3
To derive the CDF for the New Odd Weibull Inverse Rayleigh
(NOWIR) distribution, one can insert equation (4.7) into equation (2.7) in
the following manner:
𝑐 𝑐 𝑏
− 2 −
(−𝑎[−(𝑒 𝑥 ).𝑙𝑜𝑔(1−𝑒 𝑥2 )] ) (4.3)
𝐹𝑁𝑂𝑊𝐼𝑅 (𝑥, 𝑎, 𝑏, 𝑐 ) = 1 − 𝑒
By using the aforementioned equation, we derive the pdf for the
NOWIR distribution in the following manner:
𝑓𝑁𝑂𝑊𝐼𝑅 (𝑥 ) =
2𝑐
− 𝑐 𝑐
𝑒 𝑥2 − 2 − 2 𝑐 𝑐 𝑏
2𝑎𝑏𝑐[ 𝑐 −𝑒 𝑥 𝑙𝑜𝑔(1−𝑒 𝑥 )] − 2 −
𝑥 ).𝑙𝑜𝑔(1−𝑒 𝑥2 )] ) (4.4)
− (−𝑎[−(𝑒
1−𝑒 𝑥2
𝑐 𝑐 1−𝑏 𝑒
− −
𝑥 3 [−(𝑒 𝑥2 ).𝑙𝑜𝑔(1−𝑒 𝑥2 )]

where that 𝑎, 𝑏, and 𝑐 are shape parameters for the new distribution
with values greater than or equal to zero.
4.3 Neutrosophic New Odd Weibull Inverse Rayleigh
distribution
Consider the neutrosophic New Odd Inverse Rayleigh random
variable, denoted as 𝑋𝑁𝑒 , which can be expressed as 𝑋𝑁𝑒 = 𝑑 + 𝑡𝐼, where
𝑑 represents the determined component and 𝑡𝐼 represents the
indeterminate part. The values of 𝑡𝐼 lie within the interval [𝑋𝐿 , 𝑋𝑈 ], where
𝑋𝐿 and 𝑋𝑈 are the bottom and higher values of the random variable.
Similarly, the values of 𝑡𝐼 are within the interval [𝐼𝐿 , 𝐼𝑈 ]. It is important to
observe that the neutrosophic New Odd Weibull Inverse Rayleigh
distribution (NeNOWIR) becomes the classical New Odd Weibull
Inverse Rayleigh distribution when the lower bound (𝑋𝐿 ) is equal to the
upper bound (𝑋𝑈 ). The Neutrosophic cumulative density (NeCDF) of

67
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

NeNOWIR is characterized by Neutrosophic shape parameters 𝑎𝑁𝑒 ∈

[𝑎𝐿 , 𝑎𝑈 ], 𝑏𝑁𝑒 ∈ [𝑏𝐿 , 𝑏𝑈 ], and 𝑐𝑁𝑒 ∈ [𝑐𝐿 , 𝑐𝑈 ]. It may be expressed in the
following form:
𝑐 𝑐 𝑏𝑁𝑒
− 𝑁𝑒2 − 𝑁𝑒2
(−𝑎𝑁𝑒 [−(𝑒 𝑥𝑁𝑒 ).𝑙𝑜𝑔(1−𝑒 𝑥𝑁𝑒 )] ) (4.5)
𝐹𝑁𝑒 (𝑥𝑁𝑒 ) = 1 − 𝑒
Figure (4.1) displays many graphs of the NeCDF for the NeNOWIR
distribution with different intervals 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and 𝑐𝑁𝑒 .

Figure (4.1): NeCDF for NeNOWIR distribution

By using equation (4.11), we can determine the Nepdf function for the
NeNOWIR distribution in the following manner:
2𝑐
− 𝑁𝑒 𝑐 𝑐
𝑒 𝑥𝑁𝑒
2 − 𝑁𝑒2 − 𝑁𝑒2
2𝑎𝑁𝑒 𝑏𝑁𝑒 𝑐𝑁𝑒 [ 𝑥
−𝑒 𝑁𝑒 𝑥
𝑙𝑜𝑔(1−𝑒 𝑁𝑒 )]
𝑐
− 𝑁𝑒2
𝑥
1−𝑒 𝑁𝑒
𝑓𝑁𝑒 (𝑥𝑁𝑒 ) = 𝑐 𝑐 1−𝑏𝑁𝑒
− 𝑁𝑒2 − 𝑁𝑒2
3
𝑥𝑁𝑒 [−(𝑒 𝑥 𝑁𝑒 ).𝑙𝑜𝑔(1−𝑒 𝑥 𝑁𝑒 )] (4.6)
𝑐 𝑐 𝑏𝑁𝑒
− 𝑁𝑒2 − 𝑁𝑒2
𝑥 𝑥
(−𝑎𝑁𝑒 [−(𝑒 𝑁𝑒 ).𝑙𝑜𝑔(1−𝑒 𝑁𝑒 )] )
.𝑒

68
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Figure (4.2) displays many graphs of the Npdf for the NeNOWIR
distribution with different intervals 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and 𝑐𝑁𝑒 .

Figure (4.2): Nepdf for NeNOWIR distribution

Figure 4.1 shows the four graphs of NeCDF of the NeNOWIR

model with changes in the neutrosophic model parameters over different
periods. In each graph, we observe how the function's rate of ascent from
zero to one varies with different values. Smaller values of the parameters
make the curve initially flatter with a gradual rise, while larger values
cause the curve to rise more rapidly toward the maximum value. This
variation reflects the sensitivity of the NeCDF to parameter changes,
which is important in tuning the model to fit different data.
Figure 4.2 shows four graphs of the Nepdf function, showing that
varying the parameters affects the location and height of the peak, as well
as the shape of the tails. Smaller values make the distribution wider and

69
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

flatter, while larger values concentrate the density around a specific value
and produce thinner tails. This observation is important for understanding
how the model handles data with high or low dispersion.

4.4 Statistical Properties of The NeNOWIR Distribution

This section provides the derivation of many statistical aspects of the
NeNOWIR distribution, including reliability characteristics, moments,
and the quantile function.

4.5.1 Reliability Characteristics

The survival function for the NeNOWIR distribution can be found by
form:
𝑐 𝑐 𝑏𝑁𝑒
− 𝑁𝑒2 − 𝑁𝑒2
(−𝑎𝑁𝑒[−(𝑒 𝑥 𝑁𝑒 ).𝑙𝑜𝑔(1−𝑒 𝑥 𝑁𝑒 )] ) (4.7)
𝑆𝑁𝑒 (𝑥𝑁𝑒 ) = 𝑒
Figure (4.5) displays many graphs of the survival for the NeNOWIR
distribution with different intervals 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and cNe .

Figure (4.5): Survival for NeNOWIR distribution

70
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

The Hazard function for the NeNOWIR distribution is derived by

form:
2𝑐𝑁𝑒𝑢
− 𝑐 𝑐
𝑥
𝑒 𝑁𝑒𝑢
2 − 𝑁𝑒𝑢2 − 𝑁𝑒𝑢2
2𝑎𝑁𝑒𝑢 𝑏𝑁𝑒𝑢 𝑐𝑁𝑒𝑢 [ −𝑒 𝑥𝑁𝑒𝑢 𝑙𝑜𝑔(1−𝑒 𝑥𝑁𝑒𝑢 )]
𝑐
− 𝑁𝑒𝑢2
𝑥
1−𝑒 𝑁𝑒𝑢
ℎ𝑁𝑒 (𝑥𝑁𝑒 ) = (4.8)
𝑐 𝑐 1−𝑏𝑁𝑒𝑢
− 𝑁𝑒𝑢2 − 𝑁𝑒𝑢2
𝑥𝑁𝑒𝑢 𝑥 𝑥
3 [−(𝑒 𝑁𝑒𝑢 ).𝑙𝑜𝑔(1−𝑒 𝑁𝑒𝑢 )]

Figure (4.6) exhibits multiple graphs illustrating the hazard for the
NeNOWIR distribution, each corresponding to a distinct interval
𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and 𝑐𝑁𝑒 .

Figure (4.6): Hazard for NENOWIR distribution

Figure 4.3 shows the four graphs of the neutrosophic survival

function, illustrating the probability of a random variable remaining
above a certain value. Smaller values of parameters lead to a slower
decline in the survival function, meaning the probability remains high for
a longer period, while larger values lead to a faster decline, meaning the
probability of survival is lower for larger values. These differences are
important in applications of system reliability and survival analysis.
Figure 4.4 represents the hazard rate, i.e., the instantaneous rate of
failure or event at a given time. We observe that varying parameters

71
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

produce different patterns: some start at a low rate and then increase,
while others exhibit a near-constant or even decreasing pattern. This
reflects the model's flexibility in representing multiple types of hazard
behavior, whether increasing, decreasing, or complex.

4.5.2 Useful Representations for Nepdf and NeCDF of NENOWIR

Distribution
Due to the importance of the Nepdf and NeCDF distribution function
NeNOWIR and the difficulty of dealing with these functions, therefore,
these functions are simplified using exponential and logarithmic function
simplifications and the binomial expansion to find the distribution
properties. Accordingly, the Nepdf and NeCDF functions are simplified.
The simplification was done using the exponential, logarithmic functions
and binomial expansion, so that the final result for these functions is,
respectively:
Given the significance of the Nepdf and NeCDF functions of
NeNOWIR distribution and the challenges associated with working with
these functions, they are simplified through the utilization of exponential
and logarithmic function simplifications as well as binomial expansion.
This enables the determination of the distribution parameters. Hence, the
Nepdf and NeCDF are made less complex. The simplification was
achieved by the utilization of exponential and logarithmic functions, as
well as binomial expansion. Consequently, the ultimate outcomes for
these functions are as follows:
𝑐
−(𝑗𝑁𝑒 +2𝑖𝑁𝑒 𝑏𝑁𝑒 ) 𝑁𝑒2 (4.9)
𝐹𝑁𝑒 (𝑥𝑁𝑒 ) = 1 − 𝜓𝑒 𝑥 𝑁𝑒

(−1)𝑖𝑁𝑒+𝑏𝑁𝑒 +𝑗𝑁𝑒
where 𝜓 = ∑∞
𝑖𝑁𝑒 ,𝑗𝑁𝑒 =0 𝑑𝑖𝑏𝑁𝑒 ,𝑗𝑁𝑒 𝑎𝑁𝑒 𝑖𝑁𝑒
𝑖𝑁𝑒 !

𝑗 [𝑠(𝑖𝑁𝑒 𝑏𝑁𝑒 +1)−𝑗𝑁𝑒 ]

with 𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 , ,𝑗𝑁𝑒 = 𝑗𝑁𝑒 −1 ∑𝑠=1
𝑁𝑒
for 𝑗𝑁𝑒 ≥ 0 and
𝑠+1

𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 , ,0 =1

72
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝑐
−(𝑘𝑁𝑒 +2𝑖𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +𝑧𝑁𝑒 ) 𝑁𝑒2
𝐷𝑒 𝑥 𝑁𝑒
𝑓𝑁𝑒 (𝑥𝑁𝑒 ) = 3
𝑥𝑁𝑒
𝑐
(4.10)
−(𝑘𝑁𝑒 +2𝑗𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +1) 𝑁𝑒2
𝑀𝑒 𝑥 𝑁𝑒
− 3
𝑥𝑁𝑒
where
(−1)𝑘𝑁𝑒 +2𝑏𝑁𝑒 +𝑖𝑁𝑒 +𝑧𝑁𝑒−1
𝐷 = ∑∞
𝑖𝑁𝑒 ,𝑘𝑁𝑒 ,𝑧𝑁𝑒 =0 𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 −1,𝑘𝑁𝑒 2𝑐𝑁𝑒 𝑏𝑁𝑒
𝑖𝑁𝑒 !𝑎𝑁𝑒 −(𝑖𝑁𝑒+1)

and
∞
(−1)𝑖𝑁𝑒 +2𝑏𝑁𝑒 +𝑗𝑁𝑒 −1
𝑀= ∑ 𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 ,𝑗𝑁𝑒 2𝑐𝑁𝑒 𝑎𝑁𝑒 𝑖𝑁𝑒 +1𝑏𝑁𝑒
𝑖𝑁𝑒 !
𝑖𝑁𝑒 ,𝑗𝑁𝑒 =0

𝑘 [𝑠(𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 )−𝑘𝑁𝑒 ]

such that 𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 −1,𝑘𝑁𝑒 = 𝑘𝑁𝑒 −1 ∑𝑠=1
𝑁𝑒
for 𝑘𝑁𝑒 ≥ 0
𝑠+1
𝑗 [𝑙(𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 +1)−𝑗𝑁𝑒 ]
𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 −1,0 = 1, 𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 ,𝑗𝑁𝑒 = 𝑗𝑁𝑒 −1 ∑𝑙𝑁𝑒
𝑁𝑒
=1 𝑙𝑁𝑒 +1

for 𝑗𝑁𝑒 ≥ 0, and 𝑑𝑖𝑁𝑒 𝑏𝑁𝑒 +𝑏𝑁𝑒 ,0 = 1

𝑡
Given the need for the function 𝐹𝑁𝑒𝑁𝑒 to prove some mathematical
properties of the distribution, its equation was found in the form:
𝑡 𝑡𝑁𝑒
𝐹𝑁𝑒𝑁𝑒 (𝑥𝑁𝑒 ) = (1 − 𝐹𝑁𝑒𝑢 (𝑥𝑁𝑒 , 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , 𝑐𝑁𝑒 )) (4.11)
The above function was expanded to obtain the equation in the form:
𝑐
𝑡 −(𝑢𝑁𝑒 +2𝛽𝑁𝑒 𝑏𝑁𝑒𝑢 ) 𝑁𝑒2
𝐹𝑁𝑒𝑁𝑒 (𝑥𝑁𝑒 ) = Ψ𝑒 𝑥 𝑁𝑒 (4.12)

where
𝑡𝑁𝑒 (−1)𝛽𝑁𝑒+𝑏𝑁𝑒 +𝜃𝑁𝑒+𝑢𝑁𝑒
Ψ = ∑∞
𝜃𝑁𝑒 ,𝛽𝑁𝑒 ,𝑢𝑁𝑒 =0 (𝜃 ) 𝜃𝑁𝑒 𝛽𝑁𝑒 𝑎𝑁𝑒 𝛽𝑁𝑒 𝑑𝛽𝑁𝑒 𝑏𝑁𝑒 ,𝑢𝑁𝑒
𝑁𝑒 𝛽𝑁𝑒 !

𝑢 [𝑣(𝛽𝑁𝑒 𝑏𝑁𝑒 +1)−𝑢𝑁𝑒 ]

with 𝑑𝛽𝑁𝑒 𝑏𝑁𝑒 ,𝑢𝑁𝑒 = 𝑢𝑁𝑒 −1 ∑𝑣=1
𝑁𝑒
for 𝑢𝑁𝑒 ≥ 0 and
𝑣+1

𝑑𝛽𝑁𝑒 𝑏𝑁𝑒 ,0 = 1
4.5.3 Neutrosophic Quantile Function
The Quantile function is the mathematical operation that reverses the
effect of the NeCDF described in Equation (4.11). This tool is utilized to
compute the median, skewness, and kurtosis for distributions that exhibit

73
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

significant skewness or lack of moments. Furthermore, it aids in the

generation of stochastic numbers for data in simulation studies. The
equation is found as:
−𝑐𝑁𝑒
𝑥𝑁𝑒 =
1
(𝛽 )𝑏𝑁𝑒 (4.13)
ln 1 1 1
(−𝛽 )𝑏𝑁𝑒 − 𝑊−1 (−(−𝛽 )𝑏𝑁𝑒 exp [(−𝛽 )𝑏𝑁𝑒 ])
√ [ ]
log(1−𝑢𝑁𝑒 )
where 𝛽 =
𝑎𝑁𝑒

Table (4.1) displays the Quantile function for the NeNOWIR

distribution, showcasing intervals 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and cNe for various parameter
intervals.
Table (4.1): Quantile function values for different intervals for 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and 𝑐𝑁𝑒
(𝒂𝑵𝒆 , 𝒃𝑵𝒆 , 𝒄𝑵𝒆 )
𝑢𝑁𝑒 [1.2, 2.0], [1.4, 2.2], [1.6, 2.4], [1.7, 2.5], [1, 2.2 ],
[0.7, 1.7] , [ 0.8, 1.8], [ 0.9, 1.9], [0.3, 1.3], [0.6, 1.6],
[0.8, 1.8] [0.7, 1.7] [0.4, 1.4] [0.5, 1.5] [1.1, 2.1]
0.1 [0.6698, 1.3560] [0.6477, 1.3307] [0.5049, 1.2196] [0.3283, 1.0759] [0.7579, 1.4121]
0.2 [0.7903, 1.5105] [0.7539, 1.4753] [0.5814, 1.3465] [0.3838, 1.2038] [0.9110, 1.5732]
0.3 [0.9036, 1.6335] [0.8498, 1.5891] [0.6485, 1.4453] [0.4367, 1.3075] [1.0632, 1.7018]
0.4 [1.0263, 1.7467] [0.9495, 1.6927] [0.7161, 1.5346] [0.4952, 1.4047] [1.2379, 1.8205]
0.5 [1.1718, 1.8598] [1.0625, 1.7954] [0.7903, 1.6222] [0.5673, 1.5036] [1.4606, 1.9394]
0.6 [1.3609, 1.9806] [1.2012, 1.9039] [0.8780, 1.7141] [0.6666, 1.6111] [1.7776, 2.0667]
0.7 [1.6368, 2.1189] [1.3893, 2.0268] [0.9912, 1.8173] [0.8274, 1.7369] [2.2129,2.3044]
0.8 [2.1231, 2.2939] [1.6865, 2.1804] [1.1585, 1.9448] [1.1775, 1.9007] [2.3986,3.4456]
0.9 [2.5634, 3.4203] [2.3364, 2.4128] [1.4850, 2.1350] [2.1629,2.9846] [2.6863,8.0473]

The table shows that increasing the value of 𝑢𝑁𝑒 from 0.1 to 0.9
results in a shift in the quantile values from low to high ranges, reflecting
the cumulative nature of the distribution. For example, in the first set, at
𝑢𝑁𝑒 = 0.1, the quantiles are in the range [0.6698, 1.3560], while at

74
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝑢𝑁𝑒 = 0.9, they reach the range [2.5634, 3.4203], indicating a clear
widening of the values with increasing probability. It also appears that
varying initial values of the parameters directly affect the widening of the
intervals; some sets yield wider ranges at higher 𝑢𝑁𝑒 values, an indication
of increased dispersion in the tail of the distribution. Furthermore, the
differences between intervals at the same value of 𝑢𝑁𝑒 demonstrate the
sensitivity of the quantile function to parameter changes, reflecting the
flexibility of the NeNOWIR distribution in representing data with
different patterns of centering and spread.

4.5.4 Neutrosophic Moments

Moments are essential in identifying the statistical properties of a
probability distribution, such as the mean, variance, skewness, and
kurtosis. The equation can be used to calculate the 𝑘 𝑡ℎ moments of the
NeNOWIR distribution:
∞
𝔪 ) 𝔪
μ̀ 𝔪 𝑁𝑒 = 𝐸 [𝑋𝑁𝑒 ] = 𝐸 (𝑥𝑁𝑒 = ∫ 𝑥𝑁𝑒 𝑓𝑁𝑒 (𝑥𝑁𝑒 , 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , 𝑐𝑁𝑒 )𝑑𝑥𝑁𝑒
0
∞
𝑐𝑁𝑒
𝔪−3 −(𝑘𝑁𝑒 +2𝑖𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +𝑧𝑁𝑒 )𝑥𝑁𝑒 2
μ̀ 𝔪 𝑁𝑒 = 𝐷 ∫ 𝑥𝑁𝑒 𝑒 𝑑𝑥𝑁𝑒
0
∞
𝑐𝑁𝑒
𝔪−3 −(𝑘𝑁𝑒 +2𝑗𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +1)𝑥𝑁𝑒 2
−𝑀 ∫ 𝑥𝑁𝑒 𝑒 𝑑𝑥𝑁𝑒
0
𝑐𝑁𝑒
Let 𝒱 = (𝑘𝑁𝑒 + 2𝑖𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 𝑧𝑁𝑒 ) , then
𝑥𝑁𝑒 2

𝑐𝑁𝑒 (𝑘𝑁𝑒 +2𝑖𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +𝑧𝑁𝑒 ) √𝑐𝑁𝑒 (𝑘𝑁𝑒 +2𝑖𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +𝑧𝑁𝑒 )
𝑥𝑁𝑒 2 = , then 𝑥𝑁𝑒 = 1
𝒱 𝒱2

−√𝑐𝑁𝑒 (𝑘𝑁𝑒 +2𝑖𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +𝑧𝑁𝑒 )

then 𝑑𝑥𝑁𝑒 = 3 𝑑𝒱
2𝒱 2
𝑐𝑁𝑒
Let 𝒰 = (𝑘𝑁𝑒 + 2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1) , then
𝑥𝑁𝑒 2

𝑐𝑁𝑒 (𝑘𝑁𝑒 +2𝑗𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +1) √𝑐𝑁𝑒 (𝑘𝑁𝑒 +2𝑗𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +1)
𝑥𝑁𝑒 2 = , then 𝑥𝑁𝑒 = 1 ,
𝒰 𝒰2

−√𝑐𝑁𝑒 (𝑘𝑁𝑒 +2𝑗𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +1)

then 𝑑𝑥𝑁𝑒 = 3 𝑑𝒰
2𝒰 2

75
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝔪 ∞
𝐷 [𝑐𝑁𝑒 (𝑘𝑁𝑒 + 2𝑖𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 𝑧𝑁𝑒 )] 2 −1 𝔪
μ̀ 𝔪 𝑁𝑒 =− ∫ 𝒱 − 2 𝑒 −𝒱 𝑑𝒱
2
0

𝔪 ∞
𝑀[𝑐𝑁𝑒 (𝑘𝑁𝑒 + 2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1)] 2 −1 𝔪
+ ∫ 𝒰 − 2 𝑒 −𝒰 𝑑𝒱
2
0
𝔪
Γ (1 − )
μ̀ 𝔪 𝑁𝑒 = 2 [𝑀𝑊 𝔪
2
−1
𝔪
− 𝐷𝑇 2 −1] (4.14)
2
where 𝑊 = 𝑐𝑁𝑒 (𝑘𝑁𝑒 + 2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1), and 𝑇 = 𝑐𝑁𝑒 (𝑘𝑁𝑒 +
2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1)
The variance of the NeNOWIR is given by:
1 π 1 1 2
2 −2 −2 (4.15)
𝜎𝑁𝑒 = [𝑀 − 𝐷 ] − [𝑀𝑊 − 𝐷𝑇 ]
2 4
The Neutrosophic skewness (𝑆𝐾𝑁𝑒 ) and Neutrosophic kurtosis (𝐾𝑈𝑁𝑒 )
of NeNOWIR distribution are defined by [45], [41]:
1 1
−√π [𝑀𝑊 2 − 𝐷𝑇 2]
𝑆𝐾𝑁𝑒 = 3 (4.16)
𝑀−𝐷 2
( )
2
Γ(−1)[𝑀𝑊 − 𝐷𝑇]
𝐾𝑈𝑁𝑒 = −3 (4.17)
(𝑀 − 𝐷 )2
4
Table (4.2) displays a collection of values for the intervals of
moments, variance, skewness, and kurtosis by cases:
1. Case 1: 𝑎𝑁𝑒 = [1.1, 2.1], 𝑏𝑁𝑒 = [1.2, 1.6], 𝑐𝑁𝑒 = [0.1,1.1].
2. Case 2: 𝑎𝑁𝑒 = [1.1, 2.1], 𝑏𝑁𝑒 = [1.2, 1.6], 𝑐𝑁𝑒 = [0.2,1.2]
3. Case 3: 𝑎𝑁𝑒 = [1.1, 2.1], 𝑏𝑁𝑒 = [1.4, 1.8], 𝑐𝑁𝑒 = [0.3,1.3]
4. Case 4: 𝑎𝑁𝑒 = [1.1, 2.1], 𝑏𝑁𝑒 = [1.4, 1.8], 𝑐𝑁𝑒 = [0.4,1.4]
5. Case 5: 𝑎𝑁𝑒 = [1.3, 2.3], 𝑏𝑁𝑒 = [1.2, 1.6], 𝑐𝑁𝑒 = [0.5,1.5]
6. Case 6: 𝑎𝑁𝑒 = [1.3, 2.3], 𝑏𝑁𝑒 = [1.2, 1.6], 𝑐𝑁𝑒 = [0.6,1.6]
7. Case 7: 𝑎𝑁𝑒 = [1.3, 2.3], 𝑏𝑁𝑒 = [1.4, 1.8], 𝑐𝑁𝑒 = [0.7,1.7]

76
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

8. Case 8: 𝑎𝑁𝑒 = [1.3, 2.3], 𝑏𝑁𝑒 = [1.4, 1.8], 𝑐𝑁𝑒 = [0.8,1.8]

2
Table (4.2): Numerical value of 𝜇̀1 𝑁𝑒 , 𝜇̀ 2 𝑁𝑒 , 𝜇̀3 𝑁𝑒 , 𝜇̀ 4 𝑁𝑒 , 𝜎𝑁𝑒 , 𝑆𝐾𝑁𝑒 , and 𝐾𝑈𝑁𝑒 of the
NeNOWIR distribution
case

𝝁̀ 𝟏 𝑵𝒆 𝝁̀ 𝟐 𝑵𝒆 𝝁̀ 𝟑 𝑵𝒆 𝝁̀ 𝟒 𝑵𝒆 𝝈𝟐𝑵𝒆 𝑺𝑲𝑵𝒆 𝑲𝑼𝑵𝒆

[0.55803, [0.41170, [0.52279, [3.26273, [0.10029, [1.10768, [1.31769,

1 1.47046] 2.31169] 3.89327] 7.04167] 0.14942] 1.97901] 19.24902]
[0.78918, [0.82341, [1.47867, [8.38017, [0.163011, [1.10768, [1.31769,
2
1.53585] 2.52185] 4.43606] 13.0509] 0.20059] 1.979311] 19.24899]
[0.9408, [1.04312, [1.42878, [2.62632, [0.15107, [1.08675, [1.24870,
3
1.62863] 2.80351] 5.10138] 9.81444] 0.15798] 1.34110] 2.41366]
[1.08636, [1.39083, [2.19975, [4.66902, [0.16269, [1.08675, [1.24870,
4
1.69011] 3.01917] 5.70118] 11.38243] 0.21064] 1.34120] 2.41366]
[1.14627, [1.61936, [3.192326, [11.61683, [0.18171, [1.09987, [1.29184,
5
1.73344] 2.99874] 5.71151] 12.85061] 0.30540] 1.54914] 4.90045]
[1.25568, [0.46473, [4.19642, [13.21737, [1.11201, [1.09987, [1.29184,
6
1.73344] 3.19865] 6.29208] 18.50488] 0.19382] 13.2457] 85.6802]
[1.34484, [2.06933, [3.74678, [8.34776, [0.17859, [1.08139, [1.23190,
7 1.82453] 3.50751] 7.10368] 15.1558] 0.26072] 1.258669] 1.94943]
[1.43770, [2.36495, [4.57768, [10.90321, [0.18909, [1.08139, [1.23190,1.
8
1.87743] 3.71384] 7.73961] 16.99127] 0.29796] 1.25867] 94943]

Table 4.2 displays the numerical values of the first four moments of
the NeNOWIR distribution across eight different cases of the
neutrosophic parameter intervals. It can be seen that the median values
and first moments increase almost steadily from case 1 to 8, reflecting a
shift in the distribution toward larger values as the parameters are
adjusted toward higher ranges, indicating an expected increase in the
mean of the distribution.
Regarding the variance, there is a clear variation between cases, with
values remaining within a moderate range in most cases. However,
notable deviations appear in some scenarios, such as case 6, which
witnessed a significant increase in one of the variance limits (up to
1.11201), reflecting the possibility of highly dispersed data at those
parameters.
The skewness coefficient shows positive values in all cases, indicating
a rightward skew of the distribution. However, its intensity varies, as we

77
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

see relative stability in most cases, while clear jumps appear, as in case 6
(a maximum value of 13.2457), indicating a very long tail in that case.
The kurtosis coefficient reflects the degree of "sharpness" of the peak
or thickness of the tails in the distribution. Most values fall within a
moderate range, but there are extreme cases, such as case 6 (85.6802),
which indicates the presence of very heavy tails. These results reflect that
the NeNOWIR distribution is sensitive to parameter variations, especially
in the variance and upper moments. This gives it great flexibility in
representing data with different dispersion, skewness, and kurtosis
characteristics, which is important when choosing parameters to fine-tune
the fit of the distribution to real data.
Figure (4.5) presents 3D shapes of moments, skewness, and kurtosis.

(a)

78
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

(b)

(c)
Figure (4.5). (a) 3D plot of Variance, (b) 3D plot of Skewness, (c) 3D plot of
Kurtosis

Figure (4.5) shows a relationship between variance (measures a spread

of data), skewness (asymmetry), and kurtosis (measures a heaviness of
tails to distribution). Every figure shown above offers a clear visual of
how changes in one aspect, such that skewness, impact others.
The plot of variance, skewness, and kurtosis changes are significantly
depending on parameters for distribution. For example, higher values of
shape parameters lead to a broader variance, indicating that data is more
spread out. Meanwhile, changes in skewness and kurtosis reflect how
asymmetric data distribution becomes and the extent to which extreme
values are presented. The 3D plot provides a holistic view, allowing for
simultaneous observation of these measures.
It is essential for highlighting NeNOWIR distribution's behavior
across different conditions. It emphasizes NeNOWIR distribution's ability
to model datasets with varying characteristics, in order to get high
variance or skewed data, that are also useful for applications involving
real data that deviate from normality, for reliability or hazard assessment.
To find neutrosophic moment generating function for NeNOWIR
distribution using equations (4.20) and (2.15) to get:

79
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝔪
𝑛 Γ (1 − )
𝑦𝑁𝑒 ∞
μ̀ 𝑁𝑒𝑥 (𝑦𝑁𝑒 ) = ∑ [ 2 [𝑀𝑊 𝔪 𝔪
2 −1 − 𝐷𝑇 2 −1 ]] (4.18)
𝑁𝑒
𝑛=0 𝑛! 2

4.5.5 Neutrosophic Incomplete Moments

The formula for calculating the neutrosophic incomplete moments of a
random variable X Ne using equation (2.16) and 𝑓𝑁𝑒 (𝑥𝑁𝑒 ) from equation
(2,12) to get a form:
𝑦𝑁𝑒
𝑐𝑁𝑒
𝑟−3 −(𝑘𝑁𝑒 +2𝑖𝑁𝑒𝑏𝑁𝑒 +2𝑏𝑁𝑒 +𝑧𝑁𝑒 )𝑥𝑁𝑒 2
̀ 𝑟 (𝑦𝑁𝑒 ) = 𝐷 ∫
𝜇𝑁𝑒 𝑥𝑁𝑒 𝑒 𝑑𝑥𝑁𝑒
0
𝑦𝑁𝑒
𝑐𝑁𝑒
𝑟−3 −(𝑘𝑁𝑒 +2𝑗𝑁𝑒 𝑏𝑁𝑒 +2𝑏𝑁𝑒 +1)𝑥𝑁𝑒 2
−M ∫ 𝑥𝑁𝑒 𝑒 𝑑𝑥𝑁𝑒
0

The found form is:

𝑟 𝑟
𝑟 𝑇 𝑟 𝑊
𝑀𝑇 2−1 γ(1− , ) 𝐷𝑊 2−1 γ(1− , )
2 𝑥𝑁𝑒 2 2 𝑥𝑁𝑒 2 (4.19)
̀ 𝑟 (𝑦𝑁𝑒 ) =
𝜇𝑁𝑒 −
2 2

where 𝑊 = 𝑐𝑁𝑒 (𝑘𝑁𝑒 + 2𝑖𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 𝑧𝑁𝑒 ), and 𝑇 = 𝑐𝑁𝑒 (𝑘𝑁𝑒 +
2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1)

4.5.6 Neutrosophic Entropy

1. Neutrosophic Rényi Entropy
The Rényi's Entropy of a Neutrosophic random variable 𝑥𝑁𝑒 is
obtained by substituting Equation (4.16) into the equation above, as it is
illustrated below:
𝛾𝑁𝑒 ∞
1 𝛾𝑁𝑒 𝛿 −3
𝐼̀𝛿𝑁𝑒 (𝑥 )𝑁𝑒 = log [ ∑ (−1)𝑤𝑁 ( ) D. M ∫ 𝑥𝑁𝑁𝑒 𝑒 −ℋ 𝑑𝑥𝑁𝑒 ]
1 − 𝛿𝑁𝑒 𝑤𝑁𝑒
𝑤𝑁𝑒 =0 0

where ℋ = [𝑤𝑁𝑒 (2𝑖𝑁𝑒 𝑏𝑁𝑒 + 𝑧𝑁𝑒 − 2𝑗𝑁𝑒 𝑏𝑁𝑒 − 1) + 𝛾𝑁𝑒 (𝑘𝑁𝑒 +

𝑐𝑁𝑒
2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1)]
𝑥𝑁𝑒 2

By performing the process of integration on the equation provided

above, we obtain:

80
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝛿𝑁𝑒
∑𝛾𝑤𝑁𝑒
𝑁𝑒 =0
(−1)𝑤𝑁 +1 (𝑤𝛾𝑁𝑒 ) D. M Ψ 2 −1 ℳ
𝑁𝑒
log [ ]
2 (4.20)
𝐼̀𝛿𝑁𝑒 (𝑥 )𝑁𝑒 =
1 − 𝛿𝑁𝑒
where Ψ = 𝑐𝑁𝑒 [𝑤𝑁𝑒 (2𝑖𝑁𝑒 𝑏𝑁𝑒 + 𝑧𝑁𝑒 − 2𝑗𝑁𝑒 𝑏𝑁𝑒 − 1) + 𝛾𝑁𝑒 (𝑘𝑁𝑒 +
𝛿𝑁𝑒
2𝑗𝑁𝑒 𝑏𝑁𝑒 + 2𝑏𝑁𝑒 + 1)], ℳ = 𝛤 (1 − )
2

2. Neutrosophic Arimoto Entropy

Arimoto entropy is a precise metric that quantifies the level of
information or unpredictability present in a certain distribution. The AEN
measure is computed using the subsequent equation:
1
∞ 𝛿𝑁𝑒
𝛿𝑁𝑒
𝐴̀𝛿𝑁𝑒 (𝑥 )𝑁𝑒 = ([∫ [𝑓𝑁𝑒 (𝑥𝑁𝑒 , 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , 𝑐𝑁𝑒 )]𝛿𝑁𝑒 𝑑𝑥𝑁𝑒 ] − 1)
1 − 𝛿𝑁𝑒
0

Form Equation (4.26) we get:

1
𝛿𝑁𝑒 𝛿𝑁
𝛾
𝛿𝑁
𝛾𝑁𝑒
∑𝑤𝑁𝑒 =0(−1)𝑤𝑁 +1 (𝑤 )D.MΨ 2 −1 ℳ
𝐴̀𝛿𝑁𝑒 (𝑥 )𝑁𝑒 = ([ 𝑁𝑒 𝑁𝑒
] − 1) (4.21)
1−𝛿𝑁 2

3. Neutrosophic Havrda and Charvat Entropy

̀ ) measure is computed using the
The Havrda and Charvat (𝐻𝐶
subsequent equation:
1
∞ 𝛿𝑁𝑒
1
̀ 𝛿 (𝑥 )𝑁 =
𝐻𝐶 ([∫ [𝑓𝑁𝑒 (𝑥𝑁𝑒 )]𝛿𝑁𝑒 𝑑𝑥𝑁𝑒 ] − 1)
𝑁𝑒 𝑒
21−𝛿𝑁𝑒 − 1
0

Form Equation (4.26), we get:

1
𝛾𝑁𝑒 𝑤 +1 𝛾 𝛿
1 ∑𝑤 =0(−1) 𝑁 𝑁𝑒
(𝑤 )D.Mℳ 𝑁
̀ 𝛿 (𝑥 )𝑁 =
𝐻𝐶 ([ 𝑁𝑒
𝛿𝑁𝑒
𝑁𝑒
] − 1) (4.22)
𝑁𝑒 𝑒 21−𝛿𝑁𝑒 −1 1−
2Ψ 2

4. Neutrosophic Tsallis Entropy

The Tsallis (𝑇̀) measure is computed using the subsequent equation:

81
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

∞
1
𝑇̀𝛿𝑁𝑒 (𝑥 )𝑁𝑒 = (1 − ∫ [𝑓𝑁𝑒 (𝑥𝑁𝑒 , 𝑎𝑁𝑒 , 𝑏𝑁𝑒 , 𝑐𝑁𝑒 )]𝛿𝑁𝑒 𝑑𝑥𝑁𝑒 )
𝛿𝑁𝑒 − 1
0

Form Equation (4.26) we get:

𝛿𝑁𝑒
𝛾 𝛾𝑁𝑒 −1
1 ∑𝑤𝑁𝑒 =0(−1)𝑤𝑁 +1 (𝑤 )D.MΨ 2
𝑇̀𝛿𝑁𝑒 (𝑥 )𝑁𝑒 = (1 − 𝑁𝑒 𝑁𝑒
ℳ) (4.23)
𝛿𝑁𝑒 −1 2

4.5 Estimation
The NeNOWIR distribution parameters are computed using the
maximum likelihood estimation method. We calculate the log-likelihood
function for a random sample of data points 𝑥𝑁𝑒 1, 𝑥𝑁𝑒 2 , … , 𝑥𝑁𝑒 𝑚 . The
distribution adheres to the Nepdf of the NeNOWIR distribution.
𝑚

𝐿( 𝜃𝑁𝑒 , 𝑥𝑁𝑒 ) = ∏ 𝑓𝑁𝑒 (𝑥𝑁𝑒 )

𝑖=1
2𝑐𝑁𝑒
− 𝑐𝑁𝑒 𝑐𝑁𝑒
𝑥𝑁𝑒 𝑖 2 − −
𝑒 𝑥𝑁𝑒 𝑖 2 𝑥𝑁𝑒 𝑖 2
2𝑎𝑁𝑒 𝑏𝑁𝑒 𝑐𝑁𝑒 [ 𝑐𝑁𝑒 −𝑒 𝑙𝑜𝑔 (1 − 𝑒 )]
𝑛 −
𝑥𝑁𝑒 𝑖 2
1−𝑒
=∏ 𝑐𝑁𝑒 𝑐𝑁𝑒 1−𝑏𝑁𝑒
− −
𝑖=1 𝑥𝑁𝑒 𝑖 2 𝑥𝑁𝑒 𝑖 2
𝑥𝑁𝑒 𝑖 3 [− (𝑒 ) . 𝑙𝑜𝑔 (1 − 𝑒 )]

𝑐𝑁𝑒 𝑐 𝑏𝑁𝑒
− − 𝑁𝑒2
𝑥𝑁𝑒 𝑖 2 𝑥𝑁𝑒 𝑖
(−𝑎𝑁𝑒 [−(𝑒 ).𝑙𝑜𝑔(1−𝑒 )] )

.𝑒
where 𝜃𝑁𝑒 are the parameters of distribution. Then the normal log-
likelihood function L is obtained as:

82
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝐿 = 𝑚𝑙𝑜𝑔 (2) + 𝑚𝑙𝑜𝑔 (𝑎𝑁𝑒 ) + 𝑚𝑙𝑜𝑔(𝑏𝑁𝑒 ) + 𝑚𝑙𝑜𝑔(𝑐𝑁𝑒 )

2𝑐𝑁𝑒
−
𝑚 𝑥𝑁𝑒 𝑖 2 𝑐𝑁𝑒 𝑐𝑁𝑒
𝑒 −
𝑥𝑁𝑒𝑖 2
−
𝑥𝑁𝑒𝑖 2
+∑ log [ 𝑐 −𝑒 𝑙𝑜𝑔 (1 − 𝑒 )]
𝑖=1 − 𝑁𝑒2
𝑥𝑁𝑒 𝑖
1−𝑒
𝑚 𝑐𝑁𝑒 𝑐𝑁𝑒
− −
𝑥𝑁𝑒𝑖 2 𝑥𝑁𝑒 𝑖 2
+(𝑏𝑁𝑒 − 1) ∑ log (− (𝑒 ) . 𝑙𝑜𝑔 (1 − 𝑒 )) (4.24)
𝑖=1

𝑐𝑁𝑒 𝑐𝑁𝑒 𝑏𝑁𝑒

𝑚 − −
𝑥𝑁𝑒 𝑖 2 𝑥𝑁𝑒 𝑖 2
−𝑎𝑁𝑒 ∑ [− (𝑒 ) . 𝑙𝑜𝑔 (1 − 𝑒 )]
𝑖=1
𝑚

−3 ∑ log 𝑥𝑁𝑒 𝑖
𝑖=1

The maximum likelihood (MLE) estimation of the parameters

𝑎𝑁𝑒 , 𝑏𝑁𝑒 , and 𝑐𝑁𝑒 can be obtained by partially differentiating the above
equation with respect to the distribution parameters and solving the
𝜕𝐿 𝜕𝐿 𝜕𝐿
resulting non-linear equations for = 0, = 0, = 0. The
𝜕𝑎𝑁𝑒 𝜕𝑏𝑁𝑒 𝜕𝑐𝑁𝑒

solution could only be obtained by numerical approaches. These methods

utilize software like R, MAPLE, SAS, and other similar programs.

4.6 Simulation
Simulation plays a vital role in signal processing, advanced analysis,
and air temperature regulation. Simulation is used in signal processing to
gain insight into ultrasound data. Extreme analysis, especially in
meteorology, uses simulation techniques to evaluate extreme estimates.
Simulation helps researchers understand the processes and characteristics
of their disciplines.
The effectiveness of the NeNOWIR distribution estimators is
evaluated using five statistical methods, which are: MLE, OLS, WLSE,
ADE, and RTADE, and seven numerical optimization methods: Nelder-
Mead, BFGS, CG, L-BFGS-B, SANN, Newton-Raphson, and Powell

83
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

whose estimators represent a combination of numerical optimization

methods and statistical methods using a Monte Carlo simulation method
and program R. The study includes sample sizes of 50, 100, 150, and 200
observations. We set N = 1000 for the exact parameter values specified in
Tables (4.3) and (4.4). Estimators of the model parameters are averaged
to obtain the mean values. Then, the bias, MSE and RMSE are
determined.
The aforementioned charts, notably chart (4.3), illustrate the
uniformity of all estimators. As the size of the sample rises, the average
estimators of the parameters become closer to the actual values of the
parameters. Furthermore, the MSE, and RMSE exhibit a decreasing trend
as the sample size grows.

Table (4.3) : Monte Carlo simulations for statistical method conducted for
NeNOWIR
𝒂𝑵𝒆 = [𝟑, 𝟒], 𝒃𝑵𝒆 = [𝟏. 𝟓,𝟏. 𝟗], 𝒄𝑵𝒆 = [𝟒,𝟓]
Ess.
N Est. MLE OLSE WLSE ADE RTADE
Par.
[3.99178, [4.73862, [4.32817, [4.43993, [13.66783
𝑎̂𝑁𝑒
5.1434] 5.70737] 5.37399] 5.58633] , 16.7248]
[1.48804, [0.63429, [0.93179, [1.05760, [-24.1302,
Mean 𝑏̂𝑁𝑒
1.96946] 0.97873] 1.28977] 1.44993] -19.7191]
𝑐̂𝑁𝑒
[4.15725, [4.03414, [4.03280, [4.13927, [0.47706,
5.10668] 4.87072] 4.90882] 5.04556] 0.610528]
𝑎̂𝑁𝑒
[11.7349, [36.8207, [22.3517, [21.2578, [156.297,
20.5909] 42.3600] 31.2535] 32.8308] 213.8753]
[0.91061, [5.33052, [2.45857, [2.11033, [538.685,
MSE 𝑏̂𝑁𝑒
1.61547] 5.97056] 3.61799 3.13139] 785.7043]
[1.36546, [2.28251, [1.87948, [1.76287, [12.44346
𝑐̂𝑁𝑒
1.74361] 2.92973] 2.39856] 2.21794] , 19.2605]
50 [3.42563, [6.06801, [4.72776, [4.61062, [12.5019,
𝑎̂𝑁𝑒
4.53772] 6.50846] 5.59048] 5.72982] 14.62448]
[0.95426, [2.30879, [1.56798, [1.45269, [23.2096,
RMSE 𝑏̂𝑁𝑒
1.27101] 2.44347] 1.90210] 1.76957] 28.0304]
𝑐̂𝑁𝑒
[1.16853, [1.51079, [1.37094, [1.32773, [3.52752,
1.32045] 1.71164] 1.54873] 1.48927] 4.39614]
𝑎̂𝑁𝑒
[0.99178, [1.70737, [1.32817, [1.43993, [10.6678,
1.14341] 1.73862] 1.37399] 1.58633] 12.724]
[0.01195, [0.86570, [0.56820, [0.44239, [21.2191,
Bias 𝑏̂𝑁𝑒
0.06946] 0.92126] 0.61022] 0.45006] 26.0302]
𝑐̂𝑁𝑒
[0.10668, [0.12927, [0.03280, [0.04556, [3.52293,
0.15725] 0.03414] 0.09117] 0.13927] 4.38947]
𝑎̂𝑁𝑒
[3.33712, [3.57895, [3.44462, [3.50998, [14.0189,
4.59250] 4.85510] 4.73443] 4.82528] 17.3222]
100 Mean [1.51818, [1.15062, [1.29953, [1.33616, [-24.9924,
𝑏̂𝑁𝑒
1.91500] 1.43785] 1.60820] 1.65285] -20.2258]

84
CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

𝑐̂𝑁𝑒
[4.02112, [3.9245, [3.96565, [4.01959, [0.42871,
5.06446] 4.90217] 4.97987] 5.04276] 0.56363]
𝑎̂𝑁𝑒
[3.61256, [8.42918, [5.68920, [5.46567, [150.439,
7.32700] 16.4321] 11.5266] 10.9631] 207.4267]
[0.35229, [1.26409, [0.67892, [0.62771, [532.420,
MSE 𝑏̂𝑁𝑒
0.70861] 2.17868] 1.24940] 1.17736] 785.5484]
𝑐̂𝑁𝑒
[0.62821, [1.15557, [0.8294, [0.76376, [12.7704,
0.91991] 1.68417] 1.25063] 1.17656] 19.71894]
𝑎̂𝑁𝑒
[1.90067, [2.90330, [2.3852, [2.33787, [12.2653,
2.70684] 4.05366] 3.3950] 3.31106] 14.4023]
[0.59354, [1.12431, [0.82397, [0.79228, [23.0742,
RMSE 𝑏̂𝑁𝑒
0.84179] 1.47603] 1.11776] 1.08506] 28.02764]
𝑐̂𝑁𝑒
[0.79260, [1.07497, [1.11776, [0.87393, [3.57357,
0.95911] 1.29775] 1.11831] 1.08469] 4.4406]
[0.33712, [0.57895, [0.44462, [0.50998, [11.0189,
𝑎̂𝑁𝑒
0.59250] 0.85510] 0.73443] 0.82528] 13.32226]
[0.01818, [0.34937, [0.20046, [0.16383, [21.7258,
Bias 𝑏̂𝑁𝑒
0.01500] 0.46214] 0.29179] 0.24714] 26.8924]
𝑐̂𝑁𝑒
[0.02112, [0.07548, [0.03434, [0.01959, [3.57128,
0.06446] 0.09782] 0.03434] 0.04276] 4.43636]
[3.23308, [3.50905, [3.34529, [3.37150, [14.2598,
𝑎̂𝑁𝑒
4.38253] 4.52252] 4.42505] 4.48210] 18.02245]
[1.51730, [1.23002, [1.36650, [1.38582, [-20.5737,
Mean 𝑏̂𝑁𝑒
1.92175] 1.61482] 1.75160] 1.77652] -26.0029]
[4.0229, [3.99274, [4.00503, [4.02867, [0.41548,
𝑐̂𝑁𝑒
5.05377] 4.92003] 4.98370] 5.02388] 0.54070]
[2.25304, [5.71300, [3.40012, [3.27350, [144.753,
𝑎̂𝑁𝑒
4.20977] 9.61719] 6.23139] 5.93608] 205.894]
[0.22752, [0.78095, [0.39749, [0.39397, [524.658,
MSE 𝑏̂𝑁𝑒
0.38801] 1.2037] 0.67004] 0.62702] 797.8639]
𝑐̂𝑁𝑒
[0.41647, [0.80277, [0.55771, [0.52553, [12.8601,
0.61431] 1.20738] 0.83127] 0.78372] 19.91161]
150 [1.50101, [2.39018, [1.84394, [1.80928, [12.0313,
𝑎̂𝑁𝑒
2.05177] 3.10116] 2.49627] 2.43640] 14.34903]
[0.47699, [0.88371, [0.63047, [0.62767, [22.9054,
RMSE 𝑏̂𝑁𝑒
0.62290] 1.09713] 0.81856] 0.79184] 28.24649]
[0.64534, [0.89597, [0.74680, [0.72493, [3.58610,
𝑐̂𝑁𝑒
0.78378] 1.09881] 0.91174] 0.88527] 4.46224]
[0.23308, [0.50905, [0.34529, [0.37150, [11.2598,
𝑎̂𝑁𝑒
0.38253] 0.52252] 0.42505] 0.48210] 14.02245]
[0.01730, [0.26997, [0.13349, [0.11417, [22.0737,
Bias 𝑏̂𝑁𝑒
0.02175] 0.28517] 0.14839] 0.12347] 27.90290]
𝑐̂𝑁𝑒
[0.02293, [0.00725, [0.00503, [0.02867, [3.58451,
0.05377] 0.07996] 0.01629] 0.02388] 4.4592]
𝑎̂𝑁𝑒
[3.17876, [3.27579, [3.19610, [3.23282, [14.2322,
4.33415] 4.52640] 4.43945] 4.46910] 14.2322]
[1.50093, [1.33239, [1.41553, [1.42216, [-20.533, -
Mean 𝑏̂𝑁𝑒
1.90880] 1.64576] 1.75297] 1.77296] 26.1419]
𝑐̂𝑁𝑒
[4.01770, [3.95167, [3.97905, [4.00374, [0.401127
5.05346] 4.98833] 5.02742] 5.05069] 90.5302]
𝑎̂𝑁𝑒
[1.46962, [3.58738, [2.14730, [2.07787, [129.945,
200 3.07630] 7.00424] 4.6317] 4.4173] 207.956]
[0.14308, [0.43656, [0.22788, [0.22701, [493.358,
MSE 𝑏̂𝑁𝑒
0.28597] 0.87513] 0.48156] 0.45897] 804.2771]
𝑐̂𝑁𝑒
[0.29702, [0.58842, [0.39471, [0.37334, [12.9606,
0.45567] 0.87715] 0.61731] 0.58668] 20.002]
𝑎̂𝑁𝑒
[1.21228, [1.89404, [1.46536, [1.44148, [11.3993,
RMSE 1.75394] 2.64655] 2.15214] 2.10173] 14.4207]
𝑏̂𝑁𝑒 [0.37826, [0.66073, [0.47736, [0.47645, [22.2116,

85
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

0.53476] 0.93548] 0.69395] 0.67747] 28.3597]

𝑐̂𝑁𝑒
[0.54500, [0.76708, [0.62826, [0.61101, [3.60008,
0.67503] 0.93656] 0.78569] 0.76595] 4.4724]
𝑎̂𝑁𝑒
[0.17876, [0.27579, [0.19610, [0.23282, [11.2322,
0.33415] 0.52640] 0.43945] 0.46910] 14.1189]
[0.00093, [0.16760, [0.08446, [0.07783, [22.0337,
Bias 𝑏̂𝑁𝑒
0.00880] 0.25423] 0.14702] 0.12703] 28.04192]
𝑐̂𝑁𝑒
[0.01770, [0.04832, [0.02094, [0.00374, [3.59887,
0.05346] 0.01166] 0.02742] 0.05069] 4.46977]

Table (4.4) : Monte Carlo simulations for numerical optimization conducted for
NeNOWIR
𝒂𝑵𝒆 = [𝟐, 𝟐. 𝟓] 𝒃𝑵𝒆 = [𝟏. 𝟓, 𝟐] 𝒄𝑵𝒆 = [𝟑, 𝟑. 𝟓]
N Est. Nelder- L-BFGS- Newton-
[Link]. Mead
BFGS CG
B
SANN
Raphson
Powell

[3.2415, [3.2609, [0.9372 [3.2636, [3.5409, [3.2649, [0.93729

̂ 𝑵𝒆
𝒂 6.0146] 6.1086] ,1.0502] 6.1109] 6.4261] 6.1169] ,1.0502]
̂ 𝑵𝒆 [1.0257, [1.0201, [1.9516, [1.0193, [0.9292, [1.0189, [1.95169
Mean 𝒃 1.0830] 1.0661] 2.6143] 1.0657] 1.0137] 1.0646] , 2.6143]
[4.7442, [4.7777, [1.9810, [4.7824, [5.3500, [4.7847, [1.98102
𝒄̂𝑵𝒆 7.2664] 7.4058] 2.2248] 7.4091] 7.8707] 7.4180] , 2.2248]
[1.2415, [1.2609, [-1.5627 [1.2636, [1.5409, [1.2649, [-1.5627
̂ 𝑵𝒆
𝒂 3.5146] 3.6086] ,-0.949] 3.6109] 3.9261] 3.6169] ,-0.949]
̂ 𝑵𝒆 [-0.9169 [-0.9338 [0.4516, [-0.934, [-0.986, [-0.935 [0.45169
Bias 𝒃 ,-0.474] ,-0.479] 0.6143] -0.4806] -0.5707] ,-0.481] , 0.6143]
[1.7442, [1.7777, [-1.275, [1.7824, [2.3500, [0.8748, [-1.2751
𝒄̂𝑵𝒆 3.7664] 3.9058] -1.0188] 3.9091] 4.3707] 3.9180] ,-1.018]
50 [1.5414, [1.5898, [0.9020, [1.5967, [2.3746, [1.60, 1 [0.9020,
̂ 𝑵𝒆
𝒂 12.352] 13.022] 2.4421] 13.038] 15.414] 3.0822] 2.44213]
̂ 𝑵𝒆 [0.2248, [0.2302, [0.2040, [0.2310, [0.3257, [0.2314, [0.2040,
MSE 𝒃 0.8408] 0.8720] 0.3774] 0.8728] 0.9726] 0.8748] 0.37742]
[3.042, [3.1602, [1.0383, [3.1771, [5.5229, [3.1853, [1.0383,
𝒄̂𝑵𝒆 14.185] 15.255] 1.6260] 15.281] 19.103] 15.351] 1.62609]
[1.2415, [1.2609, [0.9497, [1.2636, [1.5409, [1.2649, [0.94976
̂ 𝑵𝒆
𝒂 3.5146] 3.6086] 1.5627] 3.6109] 3.9261] 3.6169] , 1.5627]
̂ 𝑵𝒆 [0.4742, [0.4798, [0.4516, [0.4806, [0.5707, [0.4810, [0.45169
RMSE 𝒃 0.9169] 0.9338] 0.6143] 0.9342] 0.9862] 0.9353] , 0.6143]
[1.7442, [1.7777, [1.0189, [1.7824, [2.3500, [1.7847, [1.01897
𝒄̂𝑵𝒆 3.7664] 3.9058] 1.2751] 3.9091] 4.3707] 3.9180] , 1.2751]
[1.1943, [1.1920, [0.9918, [1.1921, [1.4119, [1.1914, [0.99188
̂ 𝑵𝒆
𝒂 4.6034] 4.8321] 1.1835] 4.8343] 3.3859] 4.8369] , 1.183]
Mean [1.5262, [1.4801, [2.0353, [1.4797 [1.7858, [1.4791, [2.03534
̂ 𝑵𝒆
𝒃 1.8987] 1.9004] 2.6083] ,1.9003] 1.7886] 1.9008] , 2.6083]
[2.1672, [2.1646, [1.9556, [2.1647, [2.3833, [2.164, [1.95564
𝒄̂𝑵𝒆 5.1277] 5.3306] 2.4482] 5.3325] 4.1518] 5.3349] , 2.4482]
[-0.805, [-0.807, [-1.316 , [-0.807, [-0.5880 [-0.808, [-1.3164
̂ 𝑵𝒆
𝒂 2.1034] 2.3321] -1.0081] 2.3343] , 0.885] 2.3369] ,-1.008]
100 ̂ 𝑵𝒆 [-0.4737 [0.4004, [0.5353, [-0.5202 [-0.2113 [-0.520, [0.53534
Bias 𝒃 ,0.3987] -0.5198] 0.6083] ,0.4003] ,0.2858] 0.4008] , 0.6083]
[-0.8327 [-0.8353 [-1.0517 [-0.8352 [-0.6166 [-0.836, [-1.0517
𝒄̂𝑵𝒆 , 1.627] , 1.830] ,-1.044] , 1.832] , 0.651] 1.8349] ,-1.044]
[0.6490, [0.6528, [1.0162, [0.6526, [0.3457, [0.6538, [1.01629
̂ 𝑵𝒆
𝒂 4.4245] 5.4387] 1.7330] 5.4490] 0.7849] 5.4615] , 1.7330]
̂ 𝑵𝒆 [0.1590, [0.1603, [0.2865, [0.1603, [0.0446, [0.1606, [0.28659
MSE 𝒃 0.2244] 0.2702] 0.3701] 0.2707] 0.0816] 0.2712] , 0.3701]
[0.6935, [0.6978, [1.0906, [0.6976, [0.3802, [0.6988, [1.09067
𝒄̂𝑵𝒆 2.6496] 3.3511] 1.1062] 3.3583] 0.4248] 3.3669] ,1.1062]

86
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

[0.8056, [0.8079, [1.0081, [0.8078, [0.5880, [0.8085, [1.00811

̂ 𝑵𝒆
𝒂 2.1034] 2.3321] 1.3164] 2.3343] 0.8859] 2.3369] , 1.3164]
̂ 𝑵𝒆 [0.3987, [0.4004, [0.5353, [0.4003, [0.2113 [0.4008, [0.53534
RMSE 𝒃 0.4737] 0.5198] 0.6083] 0.5202] ,0.2858] 0.5208] , 0.6083]
0.83279 [0.8353, [1.0443, [0.8352, [0.6166, [0.836, [1.04435
𝒄̂𝑵𝒆 , 1.627] 1.8306] 1.0517] 1.8325] 0.6518] 1.8349] ,1.0517]
[1.6614, [1.7092, [1.0340, [1.7083, [1.5782, [1.7078, [1.10779
̂ 𝑵𝒆
𝒂 1.9785] 1.9812] 1.1077] 1.9812] 2.2336] 1.9789] ,1.0340]
Mean ̂ 𝑵𝒆 [1.5339, [1.5327, [2.0322, [1.5326, [1.4399, [1.5336, [2.03228
𝒃 2.3434] 2.3216] 2.6411] 2.3220] 2.3812] 2.3222] , 2.6411]
[2.9402, [2.9432, [1.9698, [2.9432, [2.7579, [2.8588, [1.96985
𝒄̂𝑵𝒆 2.8218] 2.8599] 2.3666] 2.8592] 3.1932] 2.9408] , 2.3666]
[-0.838, [-0.790, [-1.392, [-0.791, [-0.9217 [-0.792, [-1.392,-
̂ 𝑵𝒆
𝒂 -0.0214] -0.0187] -0.9659] -0.0187] ,0.2336] -0.021] 0.9659]
̂ 𝑵𝒆 [0.0339, [0.0327, [0.5322, [0.0326, [-0.060, [0.3222, [0.5322,
Bias 𝒃 0.3434] 0.3216] 0.6411] 0.3220] 0.3812] 0.0336] 0.64115]
[-0.678, [-0.640, [-1.133, [-0.640, [-0.7420 [-0.641, [-1.133,-
𝒄̂𝑵𝒆 -0.0597] -0.0567] -1.0301] -0.0567] ,0.1932] -0.059] 1.0301]
150 [0.0004, [0.0003, [0.9329, [0.0003, [0.0545, [0.0004, [0.93299
̂ 𝑵𝒆
𝒂 0.7031] 0.6252] 1.9382] 0.6266] 0.8495] 0.627] ,1.9382]
̂ 𝑵𝒆 [0.0011, [0.0010, [0.2833, [0.0010, [0.0036, [0.0011, [0.28333
MSE 𝒃 0.1179] 0.1034] 0.4110] 0.1037] 0.1453] 0.1038] , 0.4110]
[0.0035, [0.0032, [1.0612, [0.0032, [0.0373, [0.0035, [1.06120
𝒄̂𝑵𝒆 0.4598] 0.4097] 1.2845] 0.4106] 0.5506] 0.4111] ,1.2845]
[0.0214, [0.0187, [0.9659, [0.0187, [0.2336, [0.0210, [0.9659,
̂ 𝑵𝒆
𝒂 0.8385] 0.7907] 1.3922] 0.7916] 0.9217] 0.7921] 1.3922]
̂ 𝑵𝒆 [0.0339, [0.0327, [0.5322, [0.0326, [0.0600, [0.3222, [0.53228
RMSE 𝒃 0.3434] 0.3216] 0.6411] 0.3220] 0.3812] 0.0336] , 0.6411]
[0.0597, [0.0567, [1.0301, [0.0567, [0.1932, [0.0591, [1.03014
𝒄̂𝑵𝒆 0.6781] 0.6400] 1.1333] 0.6407] 0.7420] 0.6411] , 1.1333]
[1.7476, [1.7494, [1.0217, [1.7481, [1.6731, [1.7479, [1.02173
̂ 𝑵𝒆
𝒂 2.2180] 2.2229] 1.0515] 2.2225] 2.1441] 2.2225] , 1.0515]
̂ 𝑵𝒆 [1.5946, [1.5940, [2.0041, [1.5945, [1.6285, [1.5946, [2.00410
Mean 𝒃 2.0918] 2.0900] 2.6484] 2.0901] 2.1142] 2.0901] ,2.6484]
[2.7434, [2.7452, [1.9776, [2.7438, [2.6644, [2.7436, [1.97768
𝒄̂𝑵𝒆 3.2950] 3.2987] 2.3391] 3.2985] 3.2394] 3.2985] , 2.3391]
[-0.2819 [-0.277, [-1.448, [-0.2774 [-0.355, [-0.277, [-1.4484
̂ 𝑵𝒆
𝒂 ,-0.252] -0.2505] -0.9782] ,-0.251] -0.3268] -0.252] ,-0.978]
̂ 𝑵𝒆 [0.0918, [0.0900, [0.5041, [0.0901, [0.1142, [0.0901, [0.5041,
Bias 𝒃 0.0946] 0.0940] 0.6484] 0.0945] 0.1285] 0.0946] 0.6484]
[-0.2565 [-0.254, [-1.160, [-0.256, [-0.335, [-0.256, [-1.1608
𝒄̂𝑵𝒆 , -0.204] -0.2012] -1.0223] -0.2014] -0.2605] -0.201] ,-1.022]
200 [0.0636, [0.0627, [0.9570, [0.0634, [0.1068, [0.0635, [0.9570,
̂ 𝑵𝒆
𝒂 0.0794] 0.0767] 2.0980] 0.0769] 0.1266] 0.0769] 2.098]
̂ 𝑵𝒆 [0.0084, [0.0088, [0.2541, [0.0081, [0.013,0 [0.0080. [0.25412
MSE 𝒃 0.0089] 0.0081] 0.4204] 0.0089] .01653] 00896] , 0.4204]
[0.042,0 [0.0404, [1.0451, [0.0406, [0.0678, [0.0405, [1.04512
𝒄̂𝑵𝒆 .06583] 0.0649] 1.3475] 0.0656] 0.1125] 0.0657] , 1.3475]
[0.2523, [0.2505, [0.9782, [0.2518, [0.3268, [0.2520, [0.97826
̂ 𝑵𝒆
𝒂 0.2819] 0.2770] 1.4484] 0.2774] 0.3558] 0.2774] , 1.4486]
̂ 𝑵𝒆 [0.0918, [0.0900, [0.5041, [0.0901, [0.1142, [0.090,0 [0.50410
RMSE 𝒃 0.0946] 0.2546] 0.6484] 0.0945] 0.1285] .09466] , 0.6484]
[2.2229, [0.2012, [1.0223, [0.2014, [0.2605, [0.2014, [1.0223,
𝒄̂𝑵𝒆 0.2565] 0.2548] 1.1608] 0.2561] 0.3355] 0.2563] 1.16084]

Table (4.3) displays the results of a Monte Carlo simulation of the

NeNOWIR distribution using five statistical estimation methods (MLE,

87
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

OLSE, WLSE, ADE, and RTADE) with four sample sizes (50, 100, 150,
and 200). Means, bias, mean square error (MSE), and root mean square
error (RMSE) were evaluated. Increasing sample size reduces MSE and
RMSE values for all estimation methods, reflecting improved accuracy of
estimates as the sample size increases. Bias values also decrease with
increasing sample size, indicating that the estimated means converge
toward the true values of the parameters. The MLE method often exhibits
stable performance with relatively low bias and error magnitudes, while
the RTADE method tends to produce highly skewed values at small
sample sizes, especially for parameters b and c, making it less reliable in
these cases.
Table 4.4 focuses on the effect of seven numerical optimization
algorithms (Powell, Newton-Raphson, SANN, L-BFGS-B, CG, BFGS,
and Nelder-Mead) on estimating distribution parameters. The results are
also presented for the same metrics in Table 4.3. With the same sample
sizes, it is observed that gradient-based algorithms such as BFGS and L-
BFGS-B achieve more stable estimates with relatively low MSE and
RMSE values, especially at large sample sizes. Algorithms such as
SANN exhibit greater performance variability, especially at small
samples, reflecting the sensitivity of these methods to the number of
observations. Bias generally decreases significantly with increasing
sample size for all algorithms, indicating that most optimization methods
are capable of approaching the true values when sufficient data are
available.

4.7 Application
We demonstrate a pragmatic implementation that utilizes two
separate sources of data. The efficacy of the NeNOWIR distribution in
accurately fitting data is demonstrated. The application showcases the
benefits of NeNOWIR and its outstanding compatibility with the data.

88
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Table (4.5) presents a comparison between NeNOWIR and other

distributions. This operation is performed on the data that has already
been used.
Table (4.5): Comparative distributions
Distribution CDF
𝑐
− 𝑁𝑒
2
Neotrosophic Beta Inverse Rayleigh distribution (New) 𝛽 (𝑒 𝑥𝑁𝑒 , 𝑎 𝑁𝑒 , 𝑏𝑁𝑒 )

𝑐 𝑎𝑁𝑒 𝑏𝑁𝑒
Neotrosophic Kumaraswamy Inverse Rayleigh − 𝑁𝑒
2
1− 𝑥
(1 − (𝑒 𝑁𝑒 ) )
distribution (New)

Neotrosophic Exponential Generalized Exponential 𝑐 𝑎𝑁𝑒 𝑏𝑁𝑒

− 𝑁𝑒
2
(1 − ( 𝑥
𝑒 𝑁𝑒 ) )
Inverse Rayleigh distribution (New)

𝑐
− 𝑁𝑒
2
1 − Γ (−𝑏𝑁𝑒 log (1 − 𝑥
𝑒 𝑁𝑒 ) , 𝑎
Neotrosophic Log Gamma Inverse Rayleigh (New) 𝑁𝑒 )

𝑐 𝑏𝑁𝑒
−𝑎𝑁𝑒 − 𝑁𝑒
2
Neotrosophic [0,1]Truncated Exponentiated Exponential (1 − 𝑒 𝑥 𝑁𝑒 )

Inverse Rayleigh (New)

(1 − 𝑒 −𝑎𝑁𝑒 )𝑏𝑁𝑒
𝑐
− 𝑁𝑒
Neotrosophic Inverse Rayleigh distribution (New) 𝑥
𝑒 𝑁𝑒
2

 The Dataset-I
This data set covers COVID-19 statistics specific to the
Netherlands over a 30-day period, spanning from March 31st to April
30th, 2020. This dataset comprises crude mortality rates obtained from
[49]:
(14.918, 15.66390) (10.056, 11.18880) (12.274, 12.88770) (10.289, 10.80345)
(10.832, 11.37360) (7.099,7.45395) (5.928, 6.22440) (13.211, 13.87155)
(7.968, 8.36640) (7.584, 7.96320) (5.555, 5.83275) (6.027,6.32835)
(4.097, 4.30185) (3.611, 3.79155) (4.960, 5.20800) (7.498, 7.87290)
(6.940, 7.28700) (5.307,5.57235) (5.048, 5.30040) (2.857, 2.99985)
(2.254, 2.36670) (5.431, 5.70255) (4.462, 4.68510) (3.883,4.07715)
(3.461, 3.63405) (3.647, 3.82935) (1.974, 2.07270) (1.273, 1.33665)
(1.416, 1.48680) (4.235,4.44675)

This data hold the neutrosophic conditions by table (4.6).

89
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Table (4.6) Data-1 used, Truth, False, and indeterminacy values

No. intervals Truth (T) Falsity (F) Indeterminacy (I)
1 (14.918, 15.66390) 15.29095 -0.74590 -13.54505
2 (10.056, 11.18880) 10.62240 -1.13280 -8. 48960
3 (12.274, 12.88770) 12.58085 -0.61370 -11.96715
4 (10.289, 10.80345) 10.54623 -0.51445 -9.03178
5 (10.832, 11.37360) 11.10280 3.73300 -13.83580
6 (7.099,7.45395) 7.27648 -0.35495 -5.92153
7 (5.928, 6.22440) 6.07620 -5.28300 0.20680
8 (13.211, 13.87155) 13.54127 5.24300 -17.78427
9 (7.968, 8.36640) 8.16720 -0.39840 -6. 76880
10 (7.584, 7.96320) 7.77360 -0.37920 -6. 39440
11 (5.555, 5.83275) 5.69388 -0.27775 -4. 41613
12 (6.027,6.32835) 6.17768 -0.30135 -4. 87633
13 (4.097, 4.30185) 4.19943 -0.20485 -2. 99458
14 (3.611, 3.79155) 3.70128 -0.18055 -2. 52073
15 (4.960, 5.20800) 5.08400 -0.24800 -3. 83600
16 (7.498, 7.87290) 7.68545 -0.37490 -6. 31055
17 (6.940, 7.28700) 7.11350 -0.34700 -5. 76650
18 (5.307,5.57235) 5.43968 -0.26535 -4. 17433
19 (5.048, 5.30040) 5.17420 -0.25240 -3. 92180
20 (2.857, 2.99985) 2.92843 -0.14285 -1. 78558
21 (2.254, 2.36670) 2.31035 -0.11270 -1. 19765
22 (5.431, 5.70255) 5.56678 -0.27155 -4. 29523
23 (4.462, 4.68510) 4.57355 -0.22310 -3. 35045
24 (3.883,4.07715) 3.98008 -0.19415 -2. 78593
25 (3.461, 3.63405) 3.54753 -0.17305 -2. 37448
26 (3.647, 3.82935) 3.73818 -0.18235 -2. 55583
27 (1.974, 2.07270) 2.02335 -0.09870 -0. 92465
28 (1.273, 1.33665) 1.30483 -0.06365 -0. 24118
29 (1.416, 1.48680) 1.45140 -0.07080 -0. 38060
30 (4.235,4.44675) 4.34088 -0.21175 -3. 12913
Mean 6.36 −0.34 −6.02
Sd-values 3.45 2.67 4.72
Max-values 15.29095 5.24300 0.20680
Min-values 1.30483 −7.28300 −17.78427

Table 4.6 presents the actual data for the crude death rates from
COVID-19 in the Netherlands over a 30-day period, with the three
neutrosophic values: the true value (T), the error value (F), and the
indeterminacy value (I). The table shows that the true component is the
highest on average (6.36), while negative values dominate the error and
indeterminacy components, reflecting the presence of noise in the data,
but this does not negate the strength of the main signal represented by the
true component. The extreme values also reveal significant variability,
with some T values reaching as high as 15.29, while some F values are as
positive as 5.24, while I values mostly remain negative. To theoretically

90
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

evaluate these results, the data are plotted against the original data periods
and the distribution of the neutrosophic components for analysis.

Figure (4.6): Plot the intervals for data-I used

Figure (4.7): Histogram of Neutrosophic Components

91
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Figure (4.8): Neutrosophic parts and sum of Neutrosophic Components

The following table shows the descriptive statistics for the first
data sample, such as the mean, variance, standard deviation, maximum
and minimum values, as well as the range, skewness, and kurtosis.

Table (4.7): Descriptive statistics for the data-I

Var N Mean SD Median Trimmed Mad Min Max Range SK KU Se
[13.64 ,14.33]
[14.92 ,15.66]

[-0.18,-0.24]
[6.14 ,6.46]

[5.79 ,6.11]

[2.72 ,2.85]

[1.27 ,1.34]
[3.51,3.71]

[5.37,5.64]

[0.64,0.68]
[0.79, 0.8]
30
1

This comparison employs eight metrics. Four goodness-of-fit used

in this analysis are 𝐾𝑆, A, W, p-value as well as the HQIC, BIC, AIC,
and CAIC information criteria. Moreover.
Tables (4.8) to (4.10) present the NeNOWIR distribution with the
lowest values for AIC, CAIC, and BIC. They are compared to non-
overlapping distribution values. Furthermore, the goodness-of-fit used in
this analysis are 𝐾𝑆, A, W, p-value tests, offer compelling evidence that

92
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

the distribution of the NeNOWIR accurately corresponds to the observed

patterns in the data.

Table (4.8) Information criteria values for comparative neutrosophic distributions for
data-I
Dist. -L AIC CAIC BIC HQIC
[77.11592, 7 [160.2318, 1 [161.1549, 1 [164.4354, 1 [161.5766, 1
NENOWIR
8.8053] 63.610] 64.5337] 67.8142] 64.9554]
[82.05125, 8 [170.1025, 1 [171.0256, 1 [174.061, 17 [171.4473, 1
NeBEIR
3.6459] 73.291] 74.215] 7.4955] 74.6366]
[82.2612, 83. [170.5225, 1 [171.4456, 1 [174.7261, 1 [171.8672, 1
NeKuIR
85946] 73.718] 74.642] 77.9225] 75.0637]
[81.09271, 8 [168.1854, 1 [169.1085, 1 [172.389, 17 [169.5302, 1
NeEGIR
2.6858] 71.371] 72.2948] 5.5753] 72.7164]
[78.3050, 79. [162.6102, 1 [163.5332, 1 [166.8137, 1 [163.9549, 1
NeLGamIR
88524] 65.770] 66.6936] 69.9741] 67.1152]
Ne[0,1] [79.7998, 81. [165.5996, 1 [166.5227, 1 [169.8032, 1 [166.9444, 1
TEEIR 39258] 68.785] 69.7082] 72.9888] 70.1299]
[83.1178, 84. [168.2357, 1 [168.3785, 1 [169.6369, 1 [168.6839, 1
NeIR
74257] 71.485] 71.628] 72.8863] 71.9334]

Table (4.9): Evaluate goodness-of-fit statistics for comparative neutrosophic

distributions for data -I
Dist. W A 𝑲𝑺 p-value
[0.03997426, 0. [0.3026458, 0.3 [0.1040849 ,0.1 [0.8679853, 0.8
NENOWIR
04206249] 20364] 04103] 681221]
[0.1838734 ,0.1 [1.189859, 1.19 [0.1838884, 0.1 [0.2313112,0.23
NeBEIR
858605] 8529] 840329] 2076]
[0.1910729, 0.1 [1.230709,1.237 [0.1913032 ,0.1 [0.1935161, 0.1
NeKuIR
92792] 865] 91674] 952366]
[0.1471242, 0.1 [0.9784093, 0.9 [0.1573605, 0.1 [0.4053048,0.40
NeEGIR
476083] 785672] 573766] 543]
[9.704265, 9.72 [59.79726, 59.8 [0.9939481, 0.9 [1.221245e-15,
NeLGamIR
0721] 1517] 9355] 1.221245e-15]
Ne[0,1] [0.1030848, 0.1 [0.7154018, 0.7 [0.1583159 ,0.1 [0.3936865, 0.3
TEEIR 043297] 195903] 58886] 980525]
[0.1857869, 0.1 [1.200823, 1.20 [0.2177761 ,0.2 [0.09907227, 0.
NeIR
875191] 8047] 179032] 09941551]

93
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Table (4.10): parameter estimators by MLE for comparative neutrosophic

distributions for data-I
Dist. ̂𝑵
𝒂 𝒃𝑵 𝒄̂𝑵
NENOWIR [0.004348, 0.005277] [3.500543, 3.585131] [0.399792, 0.478028]
NeBEIR [6.21394, 7.998162] [0.7293842, 0.733630] [1.2074189, 1.426647]
NeKuIR [3.28155, 3.650918] [0.7364095, 0.739501] [2.8712703, 2.905147]
NeEGIR [0.7612317, 0.76585] [19.768029, 19.99667] [0.2669254, 0.305675]
NeLGamIR [22.57073, 23.522964] [3.6215282, 3.682028] [0.0503535, 0.053428]
Ne[0,1] [0.000408,0.0001130] [0.896525, 0.8951212] [6.5043895, 7.154179]
TEEIR
NeIR --- --- [11.84379, 13.06336]

Table 4.8 displays the results of informative measures such as AIC,

CAIC, BIC, and HQIC, which are used to evaluate the quality of
statistical models. It clearly shows that the proposed NeNOWIR
distribution achieves the lowest values on these measures, indicating a
superior fit to the data compared to the other models studied.
Table 4.9 focuses on the results of goodness-of-fit tests, such as
Kolmogorov-Smirnov, Anderson-Darling, and Cramer-von Mises, along
with the p-value. It appears that the NeNOWIR distribution achieves the
lowest values on these tests, with a high p-value (~0.868), supporting the
hypothesis that the data follow this distribution, while other distributions
fail to achieve a similar level of fit.
Finally, Table 4.10 shows the parameter estimates using the
maximum likelihood (MLE) method for each distribution. The results
show that the parameters of the NeNOWIR distribution are stable and
within reasonable ranges, reflecting its flexibility and efficiency in
representing data, unlike some other distributions that exhibit extreme or
unstable parameters, which may indicate their limited ability to adapt to
the shape of the data.

94
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Figure (4.9): Fitting pdfs NeNOWIR with histogram data-I

Figure (4.10): Empirical Fitted CDFs NeNOWIR with histogram data-I

Figure (4.9) shows the fitted probability density functions (pdfs) of

the NeNOWIR distribution alongside a histogram of the dataset. It

95
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

provides a visual confirmation that the NeNOWIR distribution closely

aligns with the empirical data.
Moreover, figure (4.10) presents the empirical cumulative
distribution function (CDF) for the fitted NeNOWIR distribution and
compares it with the observed dataset. The close match between the
empirical CDF and the fitted CDF visually reinforces the NENOWIR
model's accuracy in predicting the distribution of the data.

 The Dataset-II
Confidence intervals for the mortality rates of infants aged less than
five years [50]:
[31.53, 31.81] [29.33, 30.08] [27.23, 28.67] [25.09, 26.34]
[24.20, 24.88] [22.00, 23.50] [20.66, 22.09] [19.74, 20.59]
[18.57, 20.03] [18.04, 18.77] [16.89, 17.89] [15.92, 16.21]
[14.51, 15.92] [13.92, 14.71] [12.73, 14.32] [12.20, 13.35]
[11.18, 12.68 ] [10.21, 11.75] [10.12, 11.03] [9.12, 10.69]
[8.47, 9.42] [8.59, 9.28] [7.65, 9.03] [7.77, 8.59]
[7.23, 7.98] [6.81, 8.06]

The following table shows the values of three components for

neutrosophic and the verification condition of neutrosophic logic.

Table (4.11): Data used-II, Truth, False, and indeterminacy values

No. intervals Truth (T) Falsity (F) Indeterminacy (I)
1 [31.53, 31.81] 31.67 -0.42 -30.25
2 [29.33, 30.08] 29.71 -1.13 -27.58
3 [27.23, 28.67] 27.95 -2.16 -24.79
4 [25.09, 26.34] 25.72 -1.88 -22.84
5 [24.20, 24.88] 24.54 -1.02 -22.52
6 [22.00, 23.50] 22.75 -2.25 -19.50
7 [20.66, 22.09] 21.38 -2.15 -18.23
8 [19.74, 20.59] 20.17 0.74 -19.91
9 [19.74, 20.59] 19.30 -2.19 -16.11
10 [18.57, 20.03] 18.41 -1.10 -16.31
11 [18.04, 18.77] 17.39 -1.50 -14.89
12 [16.89, 17.89] 16.07 -0.44 -14.63
13 [15.92, 16.21] 15.22 -2.12 -12.10
14 [14.51, 15.92] 14.32 -1.19 -12.13
15 [13.92, 14.71] 13.53 -2.39 -10.14
16 [12.73, 14.32] 12.77 -1.72 -10.05
17 [11.18, 12.68 ] 11.93 -2.25 -8.68
18 [10.21, 11.75] 10.98 -2.31 -7.67

96
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

19 [10.12, 11.03] 10.58 -1.37 -8.21

20 [9.12, 10.69] 9.91 -2.36 -6.55
21 [8.47, 9.42] 8.95 -1.43 -6.52
22 [8.59, 9.28] 8.93 -1.03 -6.90
23 [7.65, 9.03] 8.34 -1.57 -5.77
24 [7.77, 8.59] 8.18 -1.03 -6.15
25 [7.23, 7.98] 7.61 -0.83 -5.78
26 [6.81, 8.06] 7.44 -1.58 -4.86
Mean 16.30 -1.49 -13.81
Sd-values 7.33 0.75 7.46
Max-values 31.67 0.74 -4.86
Min-values 7.44 -2.39 -30.25

Table 4.11 represents the data for Dataset-II, which are confidence
intervals for under-five mortality rates. These intervals were transformed
into three neutrosophic components: the true value (T), the false value
(F), and the indeterminacy value (I). The results show that the true
component, T, maintains the highest values compared to the other two
components, with an average of 16.30, reaching a maximum value of
31.67 in the first grade, while its lowest value was 7.44. This reflects that
the data exhibit a strong central signal that represents the core of the
observed phenomenon, namely the true mortality rate. In contrast, the
false component, F, often records negative values, with an average of
approximately -1.49, indicating a limited amount of negative variance or
errors in measurements or estimates, with the exception of one positive
case (0.74) in the eighth grade, which may reflect a slight difference in
estimation or data recording. The third component, the uncertainty value
(I), records a large negative mean (-13.81) and a minimum value of -
30.25, indicating that a significant portion of the data is affected by
ambiguity, perhaps due to overlaps in the statistical limits of confidence
intervals or sample size limitations. Statistical values attached to the
table, such as the standard deviation, confirm the presence of significant
variation among observations. The standard deviation for the true
component is 7.33 and for uncertainty is 7.46, indicating that the data are

97
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

not completely homogeneous and that there is variation across age groups
or time periods in mortality rates. These results, taken as a whole, reflect
that the data, while strong in representing the general trend, contain a
significant degree of ambiguity and a limited amount of deviations that
should be taken into account when modeling or interpreting. To further
describe the results of this table, the data are plotted as shown in the
figures [Link] following figures show the data used, the nature of its
analysis, and the parts of its analysis.

Figure (4.11): plot the intervals for data used-II

98
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Figure (4.12): Histogram of Neutrosophic Components

Figure (4.13): Neutrosophic parts and sum of Neutrosophic Components

The following table shows the descriptive statistics for the first
data sample, such as the mean, variance, standard deviation, maximum
and minimum values, as well as the range, skewness, and kurtosis.

99
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Table (4.12): Descriptive statistics for the data-II

N Medi Trim Rang
Mean SD Mad Min Max SK KU Se
an med e

[15.76, 16.76]

[14.21, 15.32]

[15.22, 16.26]

[31.53, 31.81]

[23.83, 24.72]

[-0.96, -0.93]
[7.27, 7.37]

[7.87, 8.27]

[6.81, 7.98]

[1.43, 1.45]
0.56
26

Table (4.13): Information criteria values for comparative neutrosophic distributions

for data-II
Dist. -L AIC CAIC BIC HQIC
[85.3825, 85. [176.7651, 1 [177.856, 17 [180.5394, 1 [177.852, 17
NENOWIR
51505] 77.030] 8.121] 80.8044] 8.117]
[85.88544, 8 [177.7709, 1 [178.8618, 1 [181.5452, 1 [178.8577, 1
NeBEIR
6.1715] 78.343] 79.4341] 82.1174] 79.43]
[85.87041, 8 [177.7408, 1 [178.8317, 1 [181.5151, 1 [178.8277, 1
NeKuIR
6.1616] 78.323] 79.4142] 82.0976] 79.4101]
[86.05436, 8 [178.1087, 1 [179.1996, 1 [181.883, 18 [179.1956, 1
NeEGIR
6.2898] 78.579] 79.6706] 2.354] 79.6665]
[85.61538, 8 [177.2308, 1 [178.3217, 1 [181.005, 18 [178.3176, 1
NeLGamIR
5.7940] 77.588] 78.6791] 1.3624] 78.675]
Ne[0,1] [97.85651, 1 [201.7138, 2 [202.8047, 2 [205.4881, 2 [202.8006, 2
TEEIR 00.230] 06.460] 07.5516] 10.235] 07.5475]
[92.99673, 9 [188.0893, 1 [188.256, 19 [189.3474, 1 [188.4516, 1
NeIR
6.0965] 94.308] 4.4753] 95.5668] 94.671]

Table (4.14): Evaluate goodness-of-fit statistics for comparative neutrosophic

distributions for data -II
Dist. W A 𝑲𝑺 p-value
[0.03759182, 0.0 [0.3121955, 0.34 [0.1037796, 0.10 [0.9077535, 0.91
NENOWIR
4765045] 24295] 50757 ] 49638]
[0.05527303, 0.0 [0.410337, 0.46 [0.1089016, 0.11 [0.8472903, 0.88
NeBEIR
6910471] 09812] 44618] 47754]
[0.0532228, 0.06 [0.3997369, 0.45 [0.1096534, 0.11 [0.8525853, 0.87
NeKuIR
792631] 49346] 37146 ] 99773]
[0.06132204, 0.0 [0.4438498, 0.48 [0.1064247, 0.11 [0.8635536, 0.89
NeEGIR
7259988] 06187] 21308] 99341]
[9.361754, 9.420 [52.43888, 52.45 [0.952514, 0.957
NeLGamIR 5.551115e-16
304] 018] 45]
Ne[0,1] [0.03860668, 0.0 [0.3194762, 0.35 [0.3477705, 0.39 [0.002582686, 0.
TEEIR 4895774] 13569] 37876] 0003809495]
[0.0494727, 0.06 [0.3775808, 0.43 [0.2695906, 0.35 [0.001862397, 0.
NeIR
379311] 07387] 61065] 03691643]

100
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Table (4.15): parameter estimators by MLE for comparative neutrosophic

distributions for data-II
Dist. ̂𝑵
𝒂 𝒃𝑵 𝒄̂𝑵
NENOWIR [0.020917, 0.02441268] [3.301722, 3.35602019] [9.896961, 13.683517]
NeBEIR [5.161696, 5.577228] [1.425926, 1.652308] [36.179173, 47.243942]
NeKuIR [28.280548, 31.782864] [1.467749, 1.730806] [6.086199, 7.350671]
NeEGIR [1.277864, 1.400316] [5.770410, 9.263341] [34.769957, 34.031967]
NeLGamIR [8.833474, 14.459887] [3.573936, 4.105925] [6.069781, 21.068758]
Ne[0,1] [-0.0036250, 0.041309] [0.9901781,0.9884544] [10.59774,10.6223550]
TEEIR
NeIR --- --- [63.33011, 65.75163]

As is the case in Tables (4.8), (4.9), and (4.10), tables (4.13),

(4.14), and (4.15) show the extent of the efficiency of the proposed model
and its superiority to the other models that were compared with it.
Like tables (4.8), (4.9), and (4.10), tables (4.13), (4.14), and (4.14)
extend the statistical analysis and evaluations to different datasets or
further examine subsets of data using statistical metrics. Though specific
details aren't highlighted in the snippet, the content would be a deeper
comparison of these distributions.

Figure (4.14): Fitting pdfs NeNOWIR with histogram data-II

101
 CHAPTER FOUR NEUTROSOPHIC NEW ODD WEIBULL INVERSE RAYLEIGH DISTRIBUTION

Figure (4.15): Empirical Fitted CDFs NeNOWIR with histogram data-II

Figure (4.14), represents a drawing of the pdf functions for the

proposed distribution with the histogram of data II.
For the data, as for Figure (4.15), shows the empirical NeCDF for
the fitted NeNOWIR distribution and compares it with the observed
dataset, which shows the extent to which the proposed model agrees with
the data used.

102
 CHAPTER FIVE CONCLUSIONS AND RECOMMENDATIONS

Chapter Five

Conclusions and Recommendations

5.1 Conclusions
In this Dissertation, we propose a new and more flexible
distribution family based on the Weibull distribution, which is a hybrid
distribution family for the T-X method with a hybrid composition method
that combines two T-X methods called NOWG. The parsimonious,
flexible, and efficient nature of the new proposal keeps its most attractive
feature. Throughout the work, distributional properties are developed
mathematically and then verified through rigorous simulations. The
analysis is conducted using several techniques.
The distribution of NOWIBX was the focus of this study. The
basic distribution functions were derived, as well as some mathematical
properties of the distribution were proven. Three conventional estimation
techniques were also used to derive the unknown parameters of the
NOWIBX distribution. In terms of the efficiency of estimating simulation
values, the OLSE technique performs better than other methods,
according to the results presented in the simulation section. Based on the
actual data examples, we can see that the NOWIBX distribution
outperforms all compared distributions in terms of data fit, followed by
KuIBX and BeIBX in terms of fit, but the NOWIBX distribution still
significantly outperforms the others large amount in terms of results.
Finally, we demonstrated that the NOWIBX distribution works better
when modeling and fitting data from the fields of medicine, economics,
industry, etc.
This Dissertation presents a new probabilistic model, the
NeNOWIR distribution with a random variable and three neutrosophic

103
 CHAPTER FIVE CONCLUSIONS AND RECOMMENDATIONS

parameters, specifically designed for applications according to

neutrosophic logic. To handle real-world data ambiguity, our exploration
of the neutrosophic NOWIR distribution has demonstrated its ability to
handle data ambiguity. The NOWIR distribution within neutrosophic
framework not only provides insight into reliability properties, moment
functions, incomplete moments, and entropy measures but also
demonstrates its versatility as a powerful tool for decoding complex data
containing upper and lower bound values. To verify its practical
relevance, we apply the NeNOWIR distribution to two real data sets,
putting it in direct comparison with well-known its classical distributions
incorporated in the neutrosophic principle. The analysis results clearly
demonstrate that NeNOWIR distribution was superior to classical
counterpart in capturing nuances of underlying data, indicating its
potential as superior modeling tool in various fields. Thus NeNOWIR
thus opens new perspectives for addressing complex real-data problems,
where ambiguity is often the rule rather than the exception. A study of
neutrosophic distributions opens.

5.2 Recommendations
The study recommends the following:
1. New sub-distributions of the New Odd Weibull-G (NOWG) family
should be developed using different underlying distributions as
inputs (such as Burr, Gamma, or Lindley) to enhance the family's
flexibility in representing data from multiple domains.
2. Propose applying the proposed family to other data types not
covered in the study, such as highly volatile financial data, extreme
weather data, and medical genomics data, to validate its
effectiveness across broader domains.

104
 CHAPTER FIVE CONCLUSIONS AND RECOMMENDATIONS

3. Develop a software library in R or Python that includes functions

for calculating the underlying properties of the family (PDF, CDF,
Moments, Quantiles), along with estimation and simulation tools,
to facilitate its use by researchers and practitioners.
4. Propose generalizing the mechanism for checking the degree of
neutrosophic in the data as a preliminary step before applying any
model and adopting clear quantitative criteria for this. This will
ensure the selection of the appropriate model and provide more
realistic results.
5. The study also recommends presenting distributions that deal with
fuzzy logic and comparing them with the results of neutrosophic
logic.
6. Further studies should be conducted to measure the sensitivity of
the new family's performance to the number of observations,
especially in the case of very small or very large samples, to more
precisely determine the limits of its application.

105
 REFERENCES

References

[1] N. Eugene, L. Carl and F. Famoye, "Beta-normal distribution and its

applications.," Communications in Statistics-Theory and methods,
pp. 497-512, 4 31 2002.
[2] A. Alzaatreh, C. Lee and F. Famoye, "A new method for generating
families of continuous distributions," Metron, p. 63–79, 71 2013.
[3] F. Smarandache, "A unifying field in logics. neutrosophy:
Neutrosophic probability, set and logic," Rehoboth, 1999.
[4] A. S. Nik, C. Chesneau, H. S. Bakouch and A. Asgharzadeh, "A new
truncated (0, b)-F family of lifetime distributions with an extensive
study to a submodel and reliability data ," Afrika Matematika, p. 3, 1
34 2023.
[5] G. M. Cordeiro, M. Alizadeh and E. M. M. Ortega, "The
Exponentiated Half‐Logistic Family of Distributions: Properties and
Applications ," Journal of Probability and Statistics, p. 864396, 1
2014.
[6] G. M. Cordeiro, E. M. M. Ortega and T. G. Ramires , "A new
generalized Weibull family of distributions: mathematical properties
and applications ," Journal of Statistical Distributions and
Applications, pp. 1-25, 2 2015.
[7] M. H. Tahir, M. Zubair, M. Mansoor, G. M. Cordeiro, M. Alizadeh
and G. G. Hamedani, "A new Weibull-G family of distributions ,"
Hacettepe Journal of Mathematics and statistics, pp. 629-647, 2 45
2016.
[8] J. Farrukh, M. A. Nasir, M. H. Tahir and N. H. Montazeri, "The odd
Burr-III family of distributions ," Journal of Statistics Applications

106
 REFERENCES

and Probability, pp. 105-122, 1 6 2017.

[9] M. Alizadeh, G. M. Cordeiro, L. G. B. Pinho and I. Ghosh, "The
Gompertz-G family of distributions ," Journal of statistical theory
and practice, pp. 179-207, 11 2017.
[10] A. D. C. Nascimento, K. F. Silva, G. M. Cordeiro, M. Alizadeh, H.
M. Yousof and G. G. Hamedani, "The odd Nadarajah-Haghighi
family of distributions: properties and applications ," Studia
Scientiarum Mathematicarum Hungarica, pp. 185-210, 2 56 2019.
[11] F. A. Bhatti, G. G. Hamedani, M. C. Korkmaz, G. M. Cordeiro, H.
M. Yousof and M. Ahmad , "On Burr III Marshal Olkin family:
development, properties, characterizations and applications ," Journal
of Statistical Distributions and Applications, pp. 1-21, 6 2019.
[12] M. S. Eliwa, Z. A. Alhussain and M. El-Morshedy, "Discrete
Gompertz-G family of distributions for over-and under-dispersed
data with properties, estimation, and applications ," Mathematics , p.
358, 3 8 2020.
[13] A. S. Hassan, E. A. El-Sherpieny and S. A. El-Taweel, "New Topp
Leone-G family with mathematical properties and applications ,"
Journal of Physics: Conference Series, p. IOP Publishing, 1 1860
2021.
[14] M. Alizadeh, A. Z. Afify, M. S. Eliwa and S. Ali , "The odd log-
logistic Lindley-G family of distributions: properties, Bayesian and
non-Bayesian estimation with applications ," Computational
statistics, pp. 281-308, 1 35 2020.
[15] Z. Shah, D. M. Khan, Z. Khan, N. Faiz, S. Hussain, A. Anwar, T.
Ahmad and K.-I. Kim, "A new generalized logarithmic–X family of
distributions with biomedical data analysis ," Applied Sciences, p.

107
 REFERENCES

3668, 6 13 2023.
[16] G. A. Mahdi, M. A. Khaleel, A. M. Gemeay, M. Nagy , A. H. Mansi,
M. M. Hossain and E. Hussam, "A new hybrid odd exponential-Φ
family: Properties and applications ," AIP Advances, 4 14 2024.
[17] H. Akaike , "Factor analysis and AIC," psychometrika, pp. 317-332,
3 52 1987.
[18] J. E. Cavanaugh , "Unifying the derivations for the Akaike and
corrected Akaike information criteria," Statistics and Probability
Letters, pp. 201-208, 2 33 1997.
[19] Y. B. Maïnassara and C. K. Célestin, "Modified Schwarz and
Hannan–Quinn information criteria for weak VARMA models,"
Statistical Inference for Stochastic Processes, pp. 199-217, 19 2016.
[20] D. A. Darling, "The kolmogorov-smirnov, cramer-von mises tests.,"
The annals of mathematical statistics, pp. 823-838, 1957.
[21] J. D. Gibbons and J. W. Pratt, "P-values: interpretation and
methodology," he American Statistician, pp. 20-25, 1 29 1975.
[22] F. Smarandache, "A geometric interpretation of the neutrosophic set-
A generalization of the intuitionistic fuzzy set," arXiv preprint math,
p. 0404520, 2004.
[23] K. Georgiev, ""A simplification of the neutrosophic sets.
Neutrosophic logic and intuitionistic fuzzy sets," Notes on
Intuitionistic Fuzzy Sets, pp. 28-31, 2 11 2005.
[24] S. K. Patro and F. Smarandache, "e. "The neutrosophic statistical
distribution, more problems, more solutions ," Infinite Study, 2016.
[25] M. B. Zeina and A. Hatip, "Neutrosophic random variables ,"
Neutrosophic Sets and Systems, pp. 44-52, 39 2021.
[26] M. M. Alanaz, M. Y. Mustafa and Z. Y. Algamal, "Neutrosophic

108
 REFERENCES

Lindley distribution with application for Alloying Metal Melting

Point," International Journal of Neutrosophic Science, pp. 65-71, 4
21 2023.
[27] M. M. Alanaz and Z. Y. Algamal, "Neutrosophic exponentiated
inverse Rayleigh distribution: Properties and Applications,"
International Journal of Neutrosophic Science, pp. 36-43, 4 21 2023.
[28] M. Y. Mustafa and Z. Y. Algamal, "Neutrosophic inverse power
Lindley distribution: A modeling and application for bladder cancer
patients," International Journal of Neutrosophic Science, pp. 216-
223, 2 21 2023.
[29] N. M. Hammood, N. K. Rashad and Z. Y. Algamal, "Neutrosophic
Topp-Leone Extended Exponential distribution modeling with
application for bladder cancer patients ," International Journal of
Neutrosophic Science, pp. 239-245, 1 25 2025.
[30] Z. Y. Algama, N. N. Alobaidi, A. A. Hamad, M. M. Alanaz and M.
Y. Mustafa, "Neutrosophic Beta-Lindley distribution: Mathematical
properties and modeling bladder cancer data," International Journal
of Neutrosophic Science, pp. 186-194, 2 23 2024.
[31] C. Alexander, G. M. Cordeiro and E. M. Ortega, "Generalized beta-
generated distributions," distributions. Computational Statistics &
Data Analysis, pp. 1880-1897, 6 56 2012.
[32] K. Zografos and N. Balakrishnan, "On families of beta-and
generalized gamma-generated distributions and associated
inference," Statistical methodology, pp. 344-362, 4 6 2009.
[33] M. M. Ristić and N. Balakrishnan, "The gamma-exponentiated
exponential distribution," Journal of statistical computation and
simulation, pp. 1191-1206, 8 82 2012.

109
 REFERENCES

[34] A. Alzaghal, A., C. Lee and F. Famoye, "Exponentiated weibull-

exponential distribution: properties and applications," Journal of
Applied Statistical Science, p. 113, 1 21 2013.
[35] H. Torabi and N. H. Montazeri, "The logistic-uniform distribution
and its applications," Communications in Statistics-Simulation and
Computation, pp. 2551-2569, 10 43 2014.
[36] M. H. Tahir, G. M. Cordeiro, M. Alizadeh, M. Mansoor, M. Zubair
and G. G. Hamedani , "The odd generalized exponential family of
distributions with applications ," Journal of Statistical Distributions
and Applications , pp. 1-28, 2 2015.
[37] Z. Ahmad, M. Elgarhy and G. G. Hamedani , "A new Weibull-X
family of distributions: properties, characterizations and
applications ," Journal of Statistical Distributions and Applications,
p. 5, 1 5 2018.
[38] A. A. Al-Babtain, M. K. Shakhatreh, M. Nassar and A. Z. Afify, "A
new modified Kies family: Properties, estimation under complete
and type-II censored samples ," Mathematics and engineering
applications, p. 1345, 8 8 2020.
[39] A. A. Al-Babtain, I. Elbatal, H. Al-Mofleh, A. M. Gemeay, A. Z.
Afify and A. M. Sarg, "The flexible burr XG family: properties,
inference, and applications in engineering science ," Symmetry, p.
474, 3 13 2021.
[40] N. A. Noori, A. A. Khalaf and M. A. Khaleel, "A New Generalized
Family of Odd Lomax-G Distributions Properties and Applications,"
Advances in the Theory of Nonlinear Analysis and Its Application,
pp. 1-16, 4 7 2023.
[41] N. A. Noori, "Exploring the Properties, Simulation, and Applications

110
 REFERENCES

of the Odd Burr XII Gompertz Distribution," Advances in the Theory

of Nonlinear Analysis and Its Application, pp. 60-75, 4 7 2023.
[42] S. NADARAJAH and S. KOTZ, "The beta exponential distribution,"
Reliability engineering & system safety, pp. 689-697, 6 91 2006.
[43] M. S. Ravikumar and R. R. Kantam, "Inverse Burr Type X
Distribution," International Journal of Advanced Engineering
Research and Applications, pp. 459-469, 11 1 2016.
[44] N. A. Noori and M. A. khaleel, "Estimation and Some Statistical
Properties of the hybrid Weibull Inverse Burr Type X Distribution
with Application to Cancer Patient Data," Iraqi Statisticians Journal,
pp. 8-29, 2 1 2024.
[45] N. A. Noori, A. A. Khalaf and M. A. Khaleel, "A new expansion of
the Inverse Weibull Distribution: Properties with Applications,"
Iraqi Statistians Journal, pp. 52-62, 1 1 2024.
[46] M. A. Khaleel, P. E. Oguntunde, J. N. Al Abbasi, N. A. Ibrahim and
M. H. AbuJarad, "The Marshall-Olkin Topp Leone-G family of
distributions: A family for generalizing probability models,"
Scientific African, p. e00470, 8 2020.
[47] E. T. Lee and J. Wang, Statistical methods for survival data analysis,
John Wiley & Sons, 2003.
[48] A. Aijaz, M. A. Lone and A. R. Aafaq, "EJAZ DISTRIBUTION A
NEW TWO PARAMETRIC DISTRIBUTION FOR MODELLING
DATA," Reliability: Theory & Applications, pp. 191-201, 1 19 2024.
[49] H. M. Almongy, E. M. Almetwally, H. M. Aljohani, A. S. Alghamdi
and E. H. Hafez, "A new extended Rayleigh distribution with
applications of COVID-19 data," Results in Physics, p. 104012, 23
2021.

111
 REFERENCES

[50] K. Zahid , M. M. A. Almazah , O. H. Odhah and H. M. Alshanbari,

"Generalized pareto model: properties and applications in
neutrosophic data modeling. Mathematical Problems in
Engineering," Mathematical Problems in Engineering, p. 3686968, 1
2022.
[51] A. M. Abd El-latif , F. A. Almulhim, N. A. Noori, M. A. Khaleel and
B. S. Alsaedi , "Properties with application to medical data for new
inverse Rayleigh distribution utilizing neutrosophic logic," Journal
of Radiation Research and Applied Sciences, p. 101391, 2 18 2025.
[52] A. R. El-Saeed, N. A. Noori, M. A. Khaleel and S. M. Alghamdi,
"Statistical properties of the Odd Lomax Burr Type X distribution
with applications to failure rate and radiation data," Journal of
Radiation Research and Applied Sciences, p. 101421, 2 18 2025.

112
 ‫المستخلص‬

‫املستخلص‬
‫تهدف هذه األطروحة إلى تقديم عائلة توزيع احتمالي جديدة تُعرف باسم "عائلة ويبل‪-‬‬
‫جي االحادية الجديدة )‪ ،" (New Odd Weibull-G Family‬وتُعد امتدادًا لتوزيع ويبل‬
‫الكالسيكي عبر دمجه بأساليب توليد هجينة إلنتاج نماذج أكثر مرونة وقدرة على تمثيل البيانات‬
‫الحقيقية‪ ،‬سواء العادية أو النتروسوفكية‪ .‬جاءت هذه الدراسة استجابةً لقصور التوزيعات التقليدية‬
‫في تمثيل البيانات غير المتماث لة أو التي تحتوي على درجة عالية من عدم اليقين‪ ،‬خاصة في‬
‫تطبيقات تحليل البقاء والبيانات الطبية والمالية‪.‬‬
‫تضمنت الدراسة تطوير نموذجي توزيع ضمن العائلة المقترحة‪ :‬أحدهما للتعامل مع‬
‫البيانات العادية‪ ،‬والثاني للتعامل مع البيانات النتروسوفكية‪ .‬جرى اشتقاق الخصائص الرياضية‬
‫األساسية للتوزيع مثل دالة الكثافة االحتمالية‪ ،‬دالة التوزيع التراكمي‪ ،‬دالة البقاء‪ ،‬والدوال العزوم‪.‬‬
‫كما تم تقدير معلمات النماذج باستخدام خمس طرق مختلفة‪ ،‬بينها اإلمكان األعظم والمربعات‬
‫الصغرى‪ ،‬وتقييم كفاءة كل طريقة من خالل محاكاة مونت كارلو‪.‬‬
‫طُبّقت النماذج على بيانات حقيقية باستخدام برنامج ‪ ،R‬وقورنت أدائها بتوزيعات أخرى‬
‫وفقًا لمعايير معلوماتية مثل معيار معلوماتية أكايكي (‪ )AIC‬ومعيار معلوماتية بيز (‪.)BIC‬‬
‫أظهرت النتائج تفوق التوزيع الجديد على تمثيل البيانات‪ ،‬خاصةً عند وجود عدم يقين أو بيانات‬
‫ذات ذيل كثيف‪ .‬كما تضمنت الدراسة أدوات للتحقق من درجة عدم اليقين في البيانات قبل‬
‫النمذجة لضمان موثوقية التحليل‪.‬‬
 ‫مجهورية العراق‬
‫وزارة التعلمي العايل والبحث العلمي‬
‫جامعة تكريت‬
‫لكية علوم احلاسوب والرايضيات‬
‫قسم الرايضيات‬
‫ادلراسات العليا‬

‫تقدير املعلمات وخصائص عائلة ‪Weibull-G‬‬

‫االحادية اجلديدة مع التطبيق‬

‫اطروحة دكتوراه مقدمة اىل‬
‫جملس لكية علوم احلاسوب والرايضيات يف جامعة تكريت‬
‫ويه جزء من متطلبات نيل شهادة دكتوراه فلسفة يف الرايضيات‬

‫من الطالب‬

‫نورالدين إياد نوري الصعب‬

‫إبرشاف‬
‫أ‪.‬د‪ .‬منذر عبداهلل خليل‬

‫‪ 1447‬هـ‬ ‫‪ 2025‬م‬