See discussions, stats, and author profiles for this publication at: [Link]

net/publication/332298495

Climate Changes Prediction Using Simple Linear Regression

Article in Journal of Computational and Theoretical Nanoscience · February 2019

DOI: 10.1166/jctn.2019.7785

CITATIONS READS

27 8,598

2 authors:

E. Sreehari G S pradeep Ghantasala

VIT University Galgotias University
10 PUBLICATIONS 76 CITATIONS 67 PUBLICATIONS 417 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by G S pradeep Ghantasala on 02 September 2020.

The user has requested enhancement of the downloaded file.

 Copyright © 2019 American Scientific Publishers Journal of
All rights reserved Computational and Theoretical Nanoscience
Printed in the United States of America Vol. 16, 1–4, 2019

Climate Changes Prediction Using Simple

Linear Regression
E. Sreehari∗ and G. S. Pradeep Ghantasala
School of Computing Science and Engineering, Galgotias University, Greater Noida 201301, Uttar Pradesh, India

The rise in global temperatures, frequent natural disasters and rising sea levels, reducing Polar
Regions have made the problem of understanding and predicting these global climate phenomena.
Prediction is a matter of prime importance and they are run as computer simulations to predict
climate variables such as temperature, precipitation, rainfall and etc. The agricultural country called
India in which 60% of the people depending upon the agriculture. Rain fall prediction is the most
important task for predicting early prediction of rainfall May helps to peasant’s as well as for the
people because most of the people in India can be depends upon the agriculture. The paper
represents simple linear regression technique for the early prediction of rainfall. It can helps to
farmers for taking appropriate decisions on crop yielding. As usually at the same time there may
be a scope to analyze the occurrence of floods or droughts. The simple linear regression analysis
methodology applied on the dataset collected over six years of Coonor in Nilagris district from Tamil
Nadu state. The experiment and our simple linear regression methodology exploit the appropriate
results for the rain fall.
Keywords: Regression, Simple Linear Regression, Prediction, Correlation Coefficient.

RESEARCH ARTICLE
1. INTRODUCTION precipitation. By this we can mitigate losses and including
Regression is the popular technique used for prediction property and damage of the public as well as for the peas-
in areas like climate prediction and other areas. The cli- ants. Weather prediction may be at regional or national
mate can be defined as the average state of atmosphere levels. Generally, two approaches used for predicting rain-
over a longer period similarly weather can be changes fall [4–6]. One is Empirical approach and other is Dynam-
strongly day by day [1]. The natural disaster which was ical approach. Empirical based on historical data to be
occurred during the year 2015 November–December south
collected and its relationship to various atmospheric vari-
India floods occurred and affected in the regions of Andhra
ables. Dynamical approach, defines physical models based
Pradesh and Tamil Nadu. In which it may results in the
loss of property approximately 100,000 cores of money, on systems of equation for prediction and can be imple-
reducing natural resources and more than 500 people were mented by using numerical rainfall forecasting method.
killed. Similarly in the year 2014 a strong tropical cyclone The most widely used empirical approaches which are
storm occurred in the regions of Vishakhapatnam and used for climate prediction [9–11], they are regression,
caused 60,000 cores of wealth damaged as well as the artificial neural network, fuzzy logic and group method
death rate 125 people were occurred. of data handling. Support vector machines are a set of
On other side most of the people in India can be supervised learning methods that create a decision maker
survived upon the cultivation. There may be a rainfall system which tries to predict new values. A simple cli-
required for doing cultivation. We are implementing the mate forecasting can be done by regression techniques.
simple regression model for predicting the future rain-
These regression techniques are discussed further. In the
fall [2–3]. By these there may be a scope to taking deci-
recent years, the use of data mining [11–13] in the field of
sion on crop yielding for the farmers. If we use the
concept of prediction there may be a scope to analyze hydrology is increasing. The studies have been performed
future occurrence of rainfall or drought, floods can be eas- using data mining process in many areas. Integrated evap-
ily determined. Generally droughts occurred due to low oration model, using DM process for three lakes in Turkey
by Keskin. Now a day’s Artificial Intelligent methods used
∗
Author to whom correspondence should be addressed. in the estimation of rainfall [14].

J. Comput. Theor. Nanosci. 2019, Vol. 16, No. xx 1546-1955/2019/16/001/004 doi:10.1166/jctn.2019.7785 1

 Climate Changes Prediction Using Simple Linear Regression Sreehari and Ghantasala

2. RELATED WORK The coefficient determination measures how well data can
Nikhil Sethi and kanwal Garg implemented a multiple lin- be represented in the regression line. It can define strength
ear regression using climate parameters by considering and direction of relationship between the dependent and
Udaipur city of Rajasthan climate data set. He can use the independent variables.
linear regression coefficient formula for calculating coeffi-
cient values. 3.1. Error Calculation
H. Hasani presented a model for forecasting the growth The error can be calculated after calculating the predicted
of height with based on polynomial regression. This model values and the difference can be calculated using the actual
can be useful for the studying growth of the children. and predicted values. The formula for calculating error is
H. Nguyen describes a model in which it can incorpo- 
rates artificial neural networks and regression for predict- yi − ȳl 2
RMSE = (3)
ing sales of industry using historical sales and economic n
indicator. Yi = Actual value, Yi = Predicted value, n = Total number
Imran Ahmad et al., presented a model for predicting of samples.
rainfall using multiple regressions in the Coonor region of RMSE can be known as root mean square error or root
Nilgris district, Tamil Nadu. He can proposed new algo- mean square deviation. It can be used measure of differ-
rithm for calculating regression coefficient values.
ence between sample and population values.
Ismail implemented the model for forecasting to pre-
dict price of gold using linear regression. Whatever he can
implement using factors inflation and money and it can 4. RAINFALL PREDICTION USING SLR
conclude that MLR perform better than prediction method The architecture shown in Figure 1 will helps you regard-
called Naïve. The forecasting strategies can be analyzed ing how to accomplish the task of rainfall prediction in a
by the data mining scientists and others [15–20]. sequential manner.
The rest of the paper is organized as follows. The Algorithm for Simple Linear Regression
In Section 2, we give a brief overview of the related work. Step 1: Function (CC, VP, A, B, X, Y, F1)
In Section 3, we explain our methodology and solution to Step 2: Collect all the data in to excel.
the climate prediction using regression [7]. In Section 4,
we define the architecture represented other several base- Data collection
lines for regression. Result can be defined in Section 5.
RESEARCH ARTICLE

In Chapter 6 we define the conclusion and future work.

Acknowledgement and references concluded at the end. Define variables in Excel

3. SIMPLE LINEAR REGRESSION

In the concept of statistical solving simple linear regres- Making excel values into array format
sion is an empirical approach and it can solves the tasks
by considering the historical data set of the climate values
or parameters. It can be only consists of single depen-
dent variable and independent variable. In the simple linear Define SLR
regression model can be exists only two variables.
The representation of simple linear regression will
be like Solve slope and intercept using coefficient formula
Y =  + X (1)
In which Y = dependent variable, X = independent vari-
able, ,  = regression coefficients.
In simple linear regression we will implement by cal- Forming SLR equation with slope and intercept
culating Slope and intercept because it will be like mathe-
matical equation of slope, intercept line. The strength and
direction of the association between the two variables can
be estimated by using the regression coefficient formula. Implement equation with present data set
Similarly there are various correlation coefficient formulas
can also be available in the mathematical and statistical
evolution processing.
The mathematical for r is given as Calculating RMSE and accuracy

xi − x̄yi − ȳ
r=  (2)
xi − x̄2 Fig. 1. Architecture to implement SLR.

2 J. Comput. Theor. Nanosci. 16, 1–4, 2019

Sreehari and Ghantasala Climate Changes Prediction Using Simple Linear Regression

Step 3: Read the values from the excel sheet f 1 = new 5. RESULTS
file (D:\\[Link]). In this section we present the experimental results for
Step 4: CC = [j] = (a. get contents ( )). the regression methodologies simple regression and multi-
Step 5: VP = [j] = (b. get contents ( )). ple regression, we have introduced briefly in the previous
Step 6: To calculate the correlation coefficient value. sections regarding regression concepts. The concept of
Step 7: Y = A + B x // here B is the correlation coefficient regression can be implemented by calculating coeffi-
value. cient, slope and the considered climate data set either
Step 8: we have to find the A = Y–Bx. // here we can get day, monthly or annual wise. As well as the per-
A value. formance of predicting the future values can be cal-
Step 9: With “A” and “B” values we can form SLR culated by using simple linear regression algorithm.
equation. By using Net beans and weka framework we imple-
mented our project. We can form the regression equation
Step 10: By using this equation with considered data set
by calculating regression coefficient will be like in this
we can obtained the predicted value which is closest to
form
dependent variable.
Y = 2035 + cloud cover−0005 (4)
Step 11: we have calculate the RMSE
Step 12: we can calculate accuracy by using confusion By substituting the considered data set in the regression
matrices. equation, we can obtain and carried out the experimental
values as shown in Table I.
The final graph representing after implementing my pro-
4.1. Initially Consideration of Requirements for SLR
posed system called simple linear regression and existing
• Data set which is in the form of climate variables with
system multiple linear regression.
numerical values. The graph we can obtained like in this way.
• Data base maintenance by Excel software. Table I represents the predicted values and the values
• JDK software. can be estimated with the help of the computed equation.
• The IDE software called Net Beans. Here the table constructed with the help of simple and
• The package of java [Link]. multiple regression strategies.
• Weka software. In Figure 2 we compare the both the regression method-
ologies simple linear regression and multiple linear regres-

RESEARCH ARTICLE
4.2. Steps Regarding the Implementation of a Project sions by considering the same data set. Here we consider
the same climate variables for predicting the future values.
• Defining the data set variables in excel.
The existing methodology called multiple linear regres-
• Installing the NetBeans IDE along with JDK if JDK is
sions can require and uses more climate parameters and
not installed in your system.
• Creation of project in NetBeans framework.
• Importing the [Link] package into the project library Table I. The results of SLR and MLR.
folder. Actual values SLR values MLR values
• Declaring the variables and source file path in the by
19.28 1924 1943
creating a file.
21.38 2028 1953
• Implementing correlation coefficient calculation using 19.24 1914 1939
programming. 20.488 20487 1841
• Define simple regression equation. 20.444 20488 2059
• Calculating slope value from using simple regression 20.487 1938 1844
equation and coefficient value.
• Final forming the predicted equation with slope and
coefficient values.
• Implementing the equation with the climate parameters
and comparing the actual values with the predicted values.
• Calculate the RMSE using past data set and obtained
values.
• Install the WEKA software.
• Obtaining confusion matrix using WEKA.
• Calculating specificity and sensitivity.
• Finally implementing the accuracy formula using speci-
ficity and sensitivity. Fig. 2. Comparing of the SLR and MLR values graph.

J. Comput. Theor. Nanosci. 16, 1–4, 2019 3

 Climate Changes Prediction Using Simple Linear Regression Sreehari and Ghantasala

it is difficult to implement this regression. In the other 2. Ozlemterzi Hindwani Monthly Rainfall Estimation by Data-Mining
way simple linear regression can be easy to implement and Process by Publishing Corporation 2012.
analyze. 3. Nikhil Sethi and KanwalGarg, 2014. Exploiting data mining tech-
nique for rainfall prediction. International Journal of Comuputer
Science and Information Technologies.
6. CONCLUSION 4. Imran Ahmed, SruthiMenon, and Nikitha, 2014. Rainfall Prediction
The natural incidents may not possible to stop and cannot Using Multiple Regression Techniques.
5. Pinky SaikiaDutta and Hitesh Tahbilder, 2013. Prediction of Rainfall
estimate in a efficient and accurate manner. In general by by Datamining Technique in Assam.
using the concept of future estimation concept or events or 6. Ismail, Z., et al., 2009. Forecasting gold pieces using multiple linear
values there may be a scope to minimize lot of problems. regression method. American Journal of Applied Sciences.
In this project we have implemented the simple regression 7. Paras, et al., 2012. A simple weather forecasting model using math-
methodology and it can predict the values in a appropriate ematical regression. Indian Research Journal of Extension Educa-
tion, 1.
manner and better than multiple linear regression. By con- 8. [Link]
sidering rainfall value as a dependent variable with other 9. Damle, C. and Yalcin, A., 2007. Journal of Hydrology
values as independent we can successfully implemented Flood Prediction Using Time Series Data Mining, 333(2–4),
the simple linear regression method. This is the concept of pp.305–316.
prediction but not in a accurate manner because we know 10. Chau, K.W. and Muttil, N., 2007. Data mining and multivariate sta-
tistical analysis for ecological system. Coastal Waters Journal of
that climate factors changes due to different reasons and Hydroinformatics, 9(4), pp.305–317.
impacts on it. 11. Roz, E.P., 2011. Water quality modeling and rainfall estimation:
A data driven approach [[Link]], University of Iowa, Iowa city,
Iowa, USA.
7. FUTURE WORK 12. Keskin, M.E. and Terziin, O., 2009. Data Mining Process for Inte-
As a future study the logistic regression we can also able grate Devoparation Model.
to implement by considering the same data set and to 13. Radhika, Y. and Shashi, M., 2009. Atmospheric Temperature Predic-
accomplish the logistic regression the dependent variable tion Using Support Vector Machines.
14. Trafalis, T.B., Richman, M.B., White, A. and Santosain, B., 2002.
must in be in binary format. Similarly we can obtain
Data Mining Techniques for Improved WSR-88D Rain Fall Estima-
results that are called predicted values and the base line of tion.
multiple regression concepts can be considered for doing 15. Ball, J.E. and Sharma, A., 2001. An Application of Artificial Neural
some mathematical calculations. So in future we can also Networks for Rainfall Forecasting.
implement the simple regression by using different cor- 16. Zhang, M., Scofield, A. and Fulcher, J., 1997. Rainfall Estimation
RESEARCH ARTICLE

relation coefficient formulas with the same methodology. Using Artificial Neural Network Group.
17. Shoji, T. and Kitaura, H., 2006. Statistical and Geostatistical Analy-
By using different correlation coefficient formulas the pre- sis of Rainfall in Central Japan.
dicted equation also changes with different values. How- 18. Ramírez, M.C.V., Velho, H.F.C. and Ferreira, N.J., 2005. Artificial
ever, by implementing this concept, we can achieve the Neural Network Technique for Rainfall Forecasting Applied to the
prediction in a better and appropriate manner. Sao Paulo Region.
19. Teegavarapu, R.S.V. and Chandramouli, V. 2005. Improved Weight-
ing Methods, Deterministic and Stochastic Data-Driven Models for
References Estimation of Missing Precipitation Records.
1. Divya Chauahan and Jawhar Thakur, 2015. Data mining techniques 20. Chiang, Y.-M., Chang, F.J., Jou, B.J.D. and Lin, P.F., 2007. Dynamic
for weather prediction. International Journal on Recent and Innova- ANN for Precipitation Estimation and Forecasting from Radar
tion Trends in Computing and Communication. observations.

Received: 24 September 2018. Accepted: 20 October 2018.

4 J. Comput. Theor. Nanosci. 16, 1–4, 2019

View publication stats