Group Members

S. No. Name Admission Number Signature

1 JUDITH AWUOR SMA/AC/00247/021

2 HERMAN MBOGO SMA/AC/00231/021

3 STANLEY SAENYI SMA/AC/00243/021

4 HILLARY OTIENO SMA/AC/00265/021

5 CLAUSE NYAOKE SMA/AC/01275/019

6 SHEILLA WANYAMA SMA/AC/00258/021

7 AUDI GEORGE SMA/AC/00260/021

8 MARK OGETO SMA/AC/00228/021

 Assessing the Impact of Lifestyle Factors on Health Insurance
Cost.

BY

group 2

A Research Project Submitted in Partial Fulfillment of the Requirements for the Award of the Degree of

Bachelor of Science in ACTURIAL SCIENCE WITH IT TOM MBOYA UNIVERSITY

MARCH 2025
Declaration

This project is our original work and therefore have not been presented for award of any degree in any other

institution and any work has been cited as required.

1. Judith Awuor

Admission Number: SMA/AC/00247/021

Signature:

2. Herman Mbogo

Admission Number: SMA/AC/00231/021

Signature:

3. Stanley Saenyi

Admission Number: SMA/AC/00243/021

Signature:

4. Hillary Otieno

Admission Number: SMA/AC/00265/021

Signature:

5. Clause Nyaoke

Admission Number: SMA/AC/01275/019

Signature:

6. Sheilla Wanyama

Admission Number: SMA/AC/00258/021

Signature:

7. Audi George

Admission Number: SMA/AC/00260/021

i
 Signature:

8. Mark Ogeto

Admission Number: SMA/AC/00228/021

Signature:

ii
Acknowledgement

We would like to express our gratitude to various individuals and groups who contributed to a research

project on Assessing te impact of Lifestyle Factors on Health Insurance [Link] guidance and support of

the supervisor ,insights from the research committee,participation of study subjects,support from colleagues

and friends ,and unwavering encouragement from family [Link] research aims to advance sustainable

agricultural develoment in the insurance sector emhazing resilience in the shaping the lifestyles.

iii
Dedication

We dedicate this project to Tom Mboya University Department of Biological And Physical Sciences and

Future Results.

iv
 Abstract

The rising cost of health insurance requires a deeper understanding of factors that influence premium

pricing and claims. Among these, lifestyle choices, including smoking, alcohol consumption, and diet, play

a critical role in determining individual health risks and insurance costs. This study aims to analyze the

impact of lifestyle factors on health insurance premiums and claims by applying actuarial and statistical

models. The research will incorporate linear regression methodology to quantify the financial effects of

these behaviors and develop a risk-based pricing framework that adjusts premiums accordingly.

The study will utilize datasets from sources such as the National Health and Nutrition Examination

Survey (NHANES), the Behavioral Risk Factor Surveillance System (BRFSS), and Kaggle medical data

repositories. Through statistical modeling, the research seeks to establish evidence-based pricing mech-

anisms that balance risk assessment with fairness in premium allocation. The findings will contribute to

actuarial science by refining risk assessment models and providing policy recommendations for insurers

and regulatory bodies like the Social Health Authority (SHA) and the National Health Insurance Fund

(NHIF). Ultimately, this research aims to support the development of a more equitable and sustainable

health insurance system that incentivizes healthier lifestyle choices while ensuring financial sustainability

for insurers.

v
Contents

1 Introduction 1

1.1 Background Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.2 Statement of the Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3 Objectives of the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3.1 Main Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.3.2 Specific Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.4 Significance of study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Literature Review 4

3 Research Methodology 9

3.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3.2 Principal Factor Analysis (PFA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.2.1 Factor Extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

3.2.2 Kaiser-Meyer-Olkin (KMO) Test and Bartlett’s Test for Sphericity . . . . . . . . . . . 10

3.2.3 Factor Loadings and Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3.2.4 Integration into Regression Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

3.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.3.1 Data sources and Sampling Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

3.3.2 Data Analysis techniques to be used . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3.3.3 Data Analysis tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4 Results and Findings 15

4.1 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

4.2 Correlation Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

vi
 4.3 Regression Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

4.4 Discussion of Key Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

5 DISCUSSION AND CONCLUSION 21

5.1 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

5.2 Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

5.3 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

5.4 Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22

5.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

References 26

Appendix 27

vii
List of Figures

1 KMO and barlett’s test. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2 data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3 Distribution of insurance premium Amount . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

4 Correlation Matrix of Key Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

5 regression analysis coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

6 dataset defination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

7 data processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

8 MLR Python Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

9 smoking status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

10 impact of exersice on premium,boxplot. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

11 bmi vs premium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

12 ECDF. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

13 heatmap exercise. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

14 line data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36

15 impact of exercise on premium. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

16 smoking status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

17 OLS Regression results. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

viii
List of Tables

1 Descriptive Statistics of Key Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2 Regression Analysis Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

ix
List of Abbreviations

• R: A Statistical Analysis Package.

• SHA: Social Health Authority.

• NHIF: National Health Insurance Fund.

• NHANES: National Health and Nutrition Examination Survey.

• BRFSS: Behavioral Risk Factor Surveillance System.

• KDHS: Kenya Demographic and Health Survey.

• MLR: Multiple Linear Regression.

• OLS: Ordinary Least Squares.

• BMI: Body Mass Index.

• ECDF: Empirical Cumulative Distribution Function Plot.

• PFA: Principal Factors Analysis.

x
Chapter 1

1 Introduction

1.1 Background Information

Health insurance premiums are determined by a variety of factors, including age, medical history, and

geographic location. However, lifestyle choices, such as smoking, alcohol consumption, physical activity, and

diet, play a significant role in influencing an individual’s health risks and, consequently, their health insurance

costs. This project aims to explore the relationship between lifestyle factors and healthcare costs and

investigate how insurers adjust premiums based on these [Link] cost of health insurance has been rising

over the years, and lifestyle choices have become a significant factor in determining the cost of premiums.

Insurers have started to incorporate lifestyle factors into their risk assessment models, but there is a lack

of comprehensive research quantifying these impacts. This lack of understanding can lead to inaccurate

risk assessments and unfair pricing [Link] investigating the correlation between lifestyle factors and

healthcare costs, we can develop better risk assessment models and fairer pricing structures. This project

will analyze the impact of lifestyle choices on health insurance premiums and explore how insurers adjust

premiums based on these factors. We will also investigate the effectiveness of incorporating lifestyle factors

into risk assessment models and explore the potential benefits and drawbacks of this [Link] project

will use a combination of quantitative and qualitative methods to collect and analyze data. We will gather

data on lifestyle choices, healthcare costs, and insurance premiums from various sources, including insurance

companies, healthcare providers, and government agencies. We will also conduct interviews with insurance

experts, healthcare providers, and policyholders to gain a deeper understanding of the relationship between

lifestyle factors and healthcare [Link] findings of this project will contribute to the existing literature on

health insurance premiums and lifestyle factors. The results will provide insights into the impact of lifestyle

choices on health insurance costs and help insurers develop better risk assessment models and fairer pricing

1
structures. The findings will also inform policymakers and stakeholders about the importance of considering

lifestyle factors in health insurance pricing and risk [Link] conclusion, this project aims to investigate

the relationship between lifestyle factors and healthcare costs and explore how insurers adjust premiums

based on these factors. By understanding the impact of lifestyle choices on health insurance premiums, we

can develop better risk assessment models and fairer pricing structures. The findings of this project will

contribute to the existing literature on health insurance premiums and lifestyle factors and provide insights

into the impact of lifestyle choices on health insurance costs.

1.2 Statement of the Problem

The rising cost of health insurance, influenced by lifestyle choices, has become a significant concern. While

some insurers have started incorporating lifestyle factors into their risk assessment models, there is a lack

of comprehensive research quantifying these impacts. This lack of understanding can lead to structures and

inaccurate risk assessments. Therefore, it is crucial to investigate the relationship between lifestyle choices

and health insurance premiums and claims to develop better risk assessment models and fairer pricing

structures.

1.3 Objectives of the Study

1.3.1 Main Objective

The main objective of this study is to Assess the impact of lifestyle factors (smoking, alcohol consump-

tion,Physical activities,exercise frequencies,income, BMI and diet) on health insurance premiums.

1.3.2 Specific Objectives

1. Analyze historical claim data to identify trends and correlations between lifestyle factors and health

risks.

2. Develop predictive models to estimate the effect of lifestyle changes on insurance claims and premiums.

2
1.4 Significance of study

To develop a risk-based pricing framework that adjusts health insurance premiums based on lifestyle-related

health risks.

This study is essential for improving the accuracy and fairness of health insurance pricing by incorporating

lifestyle factors into actuarial models. The findings will enhance risk assessment by providing insurers with

data-driven insights to refine premium calculations based on lifestyle-related health risks. By integrating

actuarial models such as credibility theory, risk scoring, and linear regression, the study will help insurers

develop more accurate methods of assessing policyholder risk based on lifestyle behaviors.

Furthermore, this research will contribute to the design of equitable premium structures, ensuring that

individuals are charged fair insurance rates that reflect their actual health risks, thus promoting sustainability

in the health insurance sector. Findings from this study will also aid policymakers, including institutions

such as the Social Health Authority (SHA) and the National Health Insurance Fund (NHIF), in formulating

regulations that incorporate lifestyle-based risk assessments into premium calculations.

By demonstrating the financial benefits of maintaining a healthy lifestyle, this study will encourage

healthier behaviors among policyholders, leading to long-term reductions in healthcare costs and claims. In

addition, the research will contribute to actuarial science by providing empirical data on the relationship

between lifestyle choices and health insurance costs, providing insights for future risk modeling advances.

Ultimately, this study seeks to bridge the gap between lifestyle-based risk assessment and health insurance

pricing, ensuring that actuarial models are both financially sustainable for insurers and fair to policyholders.

3
Chapter 2

2 Literature Review

Health insurance premiums are influenced by a variety of factors, including demographic characteristics,

health history, and lifestyle choices. The relationship between lifestyle behaviors and insurance costs has

gained significant attention in public health and economic research. Various studies have explored how factors

such as diet, physical activity, smoking, alcohol consumption, and prevalence of chronic diseases impact

premium pricing and healthcare expenditure [7] [12]. Unhealthy lifestyle choices have been strongly linked

to higher medical costs due to the increased probability of developing chronic diseases such as cardiovascular

disease, diabetes, and cancer [11]. As a result, insurers often adjust premiums based on risk assessment

models that incorporate these behavioral factors, ensuring that those who engage in high-risk behaviors

contribute more toward healthcare expenses. Although this approach promotes personal responsibility in

health management, it also raises ethical concerns about affordability and accessibility, particularly for

people with a lower socioeconomic status who may have limited resources to maintain healthier lifestyles

[9]. Furthermore, research indicates that incentives for healthier behaviors through premium reductions or

wellness programs can lead to better health outcomes and reduced long-term healthcare expenditures [10].

Many insurance providers have implemented dynamic pricing models that adjust premiums based on lifestyle

choices, rewarding individuals who participate in preventive health measures such as regular exercise, a stop

smoking and balanced nutrition [16]. For example, wearable technology and health tracking applications

are increasingly being used to monitor lifestyle factors, allowing insurers to offer personalized discounts to

policyholders demonstrating positive health habits [5]. However, concerns regarding privacy, data security,

and potential discrimination persist, as some argue that such models may unfairly penalize individuals

with genetic predispositions or conditions beyond their control [15]. Thus, while lifestyle-based premium

adjustments have the potential to promote public health and financial sustainability, their implementation

must be carefully balanced with ethical considerations and equitable access to healthcare coverage.

4
 Lifestyle choices play a crucial role in determining health outcomes, which, in turn, affect insurance

premiums. Studies have found that individuals who adopt healthier lifestyles experience lower rates of

chronic diseases such as cardiovascular conditions, diabetes, and certain cancers, reducing their overall

healthcare costs [11];[13] . Research by [12] highlights that individuals who maintain a healthy diet, engage

in regular physical activity, and avoid smoking or excessive alcohol consumption have a significantly higher

life expectancy free of major diseases.

Furthermore, the study by [14] underscores the burden of preventable diseases such as liver disease,

which is closely linked to lifestyle choices. The authors argue that policy interventions promoting healthier

behaviors could lead to reduced insurance costs by decreasing the incidence of high-cost medical conditions.

Similarly,[3] illustrate how disparities in health outcomes among COVID-19 patients were exacerbated by

pre-existing conditions related to lifestyle habits, further demonstrating the impact of individual choices on

healthcare expenses and insurance premiums.

Insurance companies incorporate lifestyle-related risks when determining premiums. According to [9],

factors such as smoking, obesity, and sedentary behavior significantly affect insurance pricing models. These

risk factors increase an individual’s likelihood of developing chronic conditions, leading to higher medi-

cal claims and consequently higher premiums. [16] further explores how premium subsidies can encourage

healthier behaviors by making insurance more accessible and rewarding risk-reducing habits. Machine learn-

ing models have been developed to predict health insurance costs based on lifestyle factors. [10] applied a

regression framework to assess how behavioral risks translate into financial implications for insurers. Their

findings suggest that incorporating lifestyle data into premium calculations enhances the accuracy of risk as-

sessments and pricing strategies. Additionally, [6] demonstrate how health insurance can influence economic

performance by mitigating financial risks associated with chronic illness and promoting preventive health-

care measures. In addition to chronic disease prevention, lifestyle choices significantly impact healthcare

utilization and long-term medical expenses, further influencing insurance premium calculations. Research

by [9] emphasizes that individuals with healthier behaviors tend to require fewer hospital visits, medical

5
procedures, and prescription medications, leading to lower overall healthcare expenditures. This reduction

in medical costs benefits both insurers and policyholders, as lower claims enable insurance providers to

maintain more stable premium rates. However, disparities in access to preventive healthcare and wellness

resources remain a challenge, as lower-income populations often face barriers to adopting healthier lifestyles

due to financial constraints, food deserts, and limited access to recreational facilities [6]. Consequently, while

lifestyle-based insurance models promote healthier living, they must also be complemented by broader public

health initiatives to ensure equitable access to health-improving resources.

Moreover, the integration of technology in health monitoring has further reinforced the link between

lifestyle factors and insurance premiums. Wearable health devices, mobile applications, and AI-driven risk

assessments enable insurers to track physical activity levels, dietary habits, and biometric data in real time

[10]. These advancements allow for personalized premium adjustments based on actual health behaviors

rather than generalized risk categorizations, potentially leading to fairer pricing structures. However, ethical

concerns regarding data privacy, potential misuse of health information, and discrimination against individ-

uals with unavoidable health risks must be carefully addressed [15]. While lifestyle-based premium models

have the potential to create financial incentives for healthier living, their implementation should be balanced

with consumer protections to prevent biases and ensure fair access to coverage for all individuals, regardless

of socioeconomic background or genetic predisposition.

The impact of lifestyle on health insurance premiums is not uniform across different socioeconomic groups.

[5] examine why informal sector workers in Indonesia struggle to pay premiums regularly, linking financial

constraints to lifestyle-related health risks. Similarly, [2] highlight that achieving universal health coverage

requires addressing disparities in access to preventive healthcare and lifestyle education. [15] discuss how high

deductibles influence healthcare-seeking behaviors, with lower-income populations often forgoing preventive

care due to cost concerns. This phenomenon exacerbates health disparities, as individuals with limited

financial resources are less likely to invest in healthy lifestyles or afford preventive screenings, leading to

higher long-term healthcare costs and insurance premiums. Moreover, socioeconomic disparities influence

6
not only access to healthcare but also the ability to maintain a healthy lifestyle, which directly impacts

insurance costs. Lower-income populations often face challenges such as food insecurity, limited access to

safe recreational spaces, and higher exposure to environmental risk factors, all of which contribute to poorer

health outcomes [6]. For example, individuals in economically disadvantaged areas may rely on inexpensive,

processed foods due to financial constraints, increasing their risk of obesity, diabetes, and cardiovascular

diseases [14] . These health conditions, in turn, result in higher insurance claims and subsequently elevated

premium rates. Thus, without policy interventions that address these structural barriers, lifestyle-based

insurance models may disproportionately penalize vulnerable populations rather than incentivizing healthier

behaviors equitably.

In addition to financial constraints, educational disparities play a significant role in shaping lifestyle

choices and health outcomes. Studies indicate that individuals with higher levels of education are more

likely to engage in preventive health measures, seek regular medical check-ups, and make informed lifestyle

decisions that reduce long-term health risks [9]. In contrast, lower levels of health literacy can lead to

delayed diagnoses, ineffective management of chronic conditions, and a greater reliance on emergency medical

services, all of which drive up insurance costs. To mitigate these disparities,[2] advocate for comprehensive

public health campaigns and government-subsidized wellness programs aimed at improving health awareness

and preventive care accessibility. By integrating such initiatives with insurance structures, policymakers can

work toward reducing the socioeconomic gap in lifestyle-related health risks and promoting a fairer, more

inclusive insurance system.

Effective policy interventions can mitigate the financial burden of unhealthy lifestyles on insurance sys-

tems. [1] argue that integrating lifestyle-based risk assessments into public health strategies can optimize

premium structures and encourage preventive care. [4] emphasize the importance of workplace wellness

programs and tax incentives to promote healthier behaviors and reduce healthcare expenditures.

In addition,[8] recommend leveraging data-driven approaches to identify high-risk individuals and tailor

insurance plans accordingly. Such initiatives could bridge the gap between lifestyle behaviors and insurance

7
affordability, ensuring equitable access to healthcare while maintaining financial viability for insurers.

Lifestyle factors significantly impact health insurance premiums by influencing disease risk and health-

care expenditures. Empirical studies underscore the necessity of incorporating lifestyle considerations into

insurance pricing models and public health policies. While healthier behaviors can lead to reduced premi-

ums and improved health outcomes, socioeconomic disparities must be addressed to ensure equitable access

to insurance benefits. Future research should explore innovative policy mechanisms that balance financial

sustainability with incentives for healthier lifestyles, ultimately enhancing both individual and population

health.

8
Chapter 3

3 Research Methodology

3.1 Methodology

This study employs multiple linear regression model to Model the relationship between lifestyle factors

(independent variables) and health insurance metrics (dependent variables).

Predicts premium changes based on individual lifestyle habits.

premium cost = β0 + β1 X1 + β2 X2 + β3 X3 + ϵ (1)

PremiumCost = β0 + β1 (Smoking) + β2 (AlcoholConsumption) + β3 (BMI) + ϵ (2)

where:

• PremiumCost = Dependent variable.

• β0 = Intercept term.

• β1 , β2 , β3 = Coefficients for lifestyle variables.

• ϵ = Error term.

We applied Ordinary Least Squares to estimate the coefficients using R

PremiumCost = β0 + β1 (Smoking) + β2 (Exercise Frequency) + ϵ (3)

where:

• PremiumCost = Dependent variable (insurance premium amount).

• β0 = Intercept term.

9
 • β1 , β2 = Coefficients for independent variables.

• Smoking = Binary variable (1 = Yes, 0 = No).

• Exercise Frequency = Ordinal variable (0 = Rarely, 1 = Monthly, 2 = Weekly, 3 = Daily).

• ϵ = Error term.

3.2 Principal Factor Analysis (PFA)

To reduce multicollinearity and identify key underlying lifestyle factors influencing health insurance pre-

miums, we applied Principal Factor Analysis (PFA). This method allows for dimensionality reduction

while preserving the variance explained by key lifestyle-related variables.

3.2.1 Factor Extraction

The following lifestyle variables were included in the factor analysis:

• Smoking Status (binary: 1 = smoker, 0 = non-smoker)

• Alcohol Consumption (ordinal: 1 = rarely, 2 = monthly, 3 = weekly, 4 = daily)

• Exercise Frequency (ordinal: 0 = rarely, 1 = monthly, 2 = weekly, 3 = daily)

• Body Mass Index (BMI) (continuous)

• Dietary Quality Score (continuous)

We used the Kaiser-Meyer-Olkin (KMO) test to assess sampling adequacy and Bartlett’s test to verify

factorability. The results confirmed that the dataset was suitable for factor analysis.

3.2.2 Kaiser-Meyer-Olkin (KMO) Test and Bartlett’s Test for Sphericity

To determine the suitability of the dataset for Principal Factor Analysis (PFA), we performed the Kaiser-

Meyer-Olkin (KMO) test and Bartlett’s test for sphericity.

10
KMO Test The KMO test measures sampling adequacy, ensuring that variables have sufficient correlation

for factor analysis. The computed KMO statistic was:

Figure 1: KMO and barlett’s test.

KM O = 0.7523 (4)

A KMO value of 0.7523 indicates that the dataset is good for factor analysis.

Bartlett’s Test for Sphericity Bartlett’s test assesses whether the correlation matrix significantly differs

from an identity matrix. The test result was:

χ2 (10) = 2305.67, p < 0.000 (5)

Since the p-value is below 0.05, we reject the null hypothesis that variables are uncorrelated. This

confirms that factor analysis is appropriate.

Implications for Factor Analysis Based on these results:

• The KMO value (0.7523) suggests that the dataset is well-suited for factor analysis.

• The significant Bartlett’s test (p < 0.000) confirms that the correlation matrix is factorable.

Thus, we proceed with Principal Factor Analysis to extract meaningful lifestyle-related factors influencing

health insurance premiums.

3.2.3 Factor Loadings and Interpretation

Using the eigenvalue criterion (λ > 1) and scree plot analysis, we retained the top two factors:

11
 1. Health Risk Factor (Factor 1): This factor had high loadings from BMI, smoking, and dietary qual-

ity, indicating that individuals with poor health behaviors were more likely to incur higher insurance

costs.

2. Activity and Lifestyle Factor (Factor 2): Exercise frequency and alcohol consumption loaded

onto this factor, suggesting that active lifestyles and moderate alcohol intake influenced premium

adjustments.

3.2.4 Integration into Regression Model

The extracted factors were used as independent variables in a multiple linear regression model:

P remiumCost = β0 + β1 (F actor1 ) + β2 (F actor2 ) + ϵ (6)

where:

• F actor1 : Health Risk Factor (higher values indicate higher risk)

• F actor2 : Activity and Lifestyle Factor

• ϵ: Error term

The results indicate that F actor1 was significantly associated with higher insurance premiums (p < 0.01),

while F actor2 had a smaller but statistically significant negative effect on premiums (p < 0.05). These

findings support the hypothesis that insurers adjust premiums based on lifestyle-related health risks.

The estimated regression equation based on the dataset is:

ˆ
PremiumCost = 1132.78 − 197.66(Smoking) − 34.69(Exercise Frequency) (7)

The statistical significance of the model is analyzed as follows:

• Smoking has a significant negative impact on premiums (p = 0.018).

12
 • Exercise Frequency is not statistically significant (p = 0.349).

• The model explains only 1.3% of the variance in premium costs (R2 = 0.013), suggesting other factors

influence the premium.

The results suggest that smoking status significantly affects insurance premiums, while exercise frequency

does not have a substantial impact. Future improvements to the model can be made by including additional

predictors such as BMI, age, and medical history.

3.3 Data Analysis

3.3.1 Data sources and Sampling Techniques

We utilized data obtained from publicly available sources, including the Kaggale software Reports,NHIF

reports and the Kenya Demographic and Health Survey (KDHS). These sources provide valuable information

on healthcare expenditures, insurance coverage, and the prevalence of lifestyle-related health conditions.

Additionally, relevant academic papers and actuarial research on healthcare funding models were reviewed

to supplement the findings.

The datasets used in this study were originally collected using stratified and cluster sampling techniques.

The KDHS employed stratified sampling to ensure representative health data across different demographic

groups, while NHIF hospital claims data follows systematic sampling approaches in some instances. Stratified

sampling ensures that key subgroups, such as different age groups or income levels, are proportionally

represented. Cluster sampling, on the other hand, is commonly applied in national health surveys, where

specific regions or counties are selected to represent the larger population.

The selection of secondary data sources allows for a broad and representative sample that aligns with

the research objectives. By utilizing existing datasets from reputable organizations, this study ensures data

accuracy, reliability, and relevance in assessing the impact of lifestyle on healthcare funding and insurance

models.

13
3.3.2 Data Analysis techniques to be used

This study employs a Multiple Linear Regression (MLR) model to analyze the impact of lifestyle factors on

healthcare expenditures under the NHIF. The MLR model is used to determine the relationship between

healthcare costs and multiple independent variables, such as Body Mass Index (BMI), age, smoking status,

and other demographic or lifestyle factors. The model estimates the effect of these variables on healthcare

costs by determining the strength and direction of their relationships while controlling for other influencing

factors.

The regression coefficients are estimated using the Ordinary Least Squares (OLS) method, which mini-

mizes errors in prediction. To evaluate the model’s performance, statistical measures such as the coefficient

of determination (R2 ) will assess how well the independent variables explain variations in healthcare costs.

Additionally, the adjusted R2 will be used to account for the number of predictors and prevent overfitting.

The overall significance of the model will be tested using the F-test, while t-tests will determine the individual

significance of each predictor variable. Residual analysis will also be conducted to verify that key regression

assumptions—such as normality, homoscedasticity, and absence of multicollinearity—are satisfied.

The interpretation of the regression results will provide insights into the impact of BMI and other lifestyle

factors on healthcare costs. This analysis will help determine which variables significantly influence medical

expenses and guide policy recommendations for SHA and NHIF. By applying MLR, this study aims to offer

data-driven insights into the relationship between lifestyle choices and healthcare expenditures, contributing

to improved healthcare funding strategies in Kenya.

3.3.3 Data Analysis tools

This study utilizes R and Python for data analysis. R is employed for statistical modeling, including Multiple

Linear Regression (MLR), hypothesis testing, and data visualization using packages such as tidyverse and

ggplot2. Python, with libraries like pandas, statsmodels, and scikit-learn, supports data preprocessing,

regression modeling, and predictive analytics. These tools ensure efficient and accurate analysis of the data.

14
Chapter 4

4 Results and Findings

This chapter presents the results of the statistical analysis conducted to examine the relationship between

lifestyle factors and health insurance premiums. The findings are organized into key subsections, including

descriptive statistics, correlation analysis, and regression modeling. Each section provides insights into

how demographic characteristics, health behaviors, and financial indicators influence insurance premium

[Link] results are interpreted in alignment with existing literature and the research objectives.

4.1 Descriptive Statistics

We computed descriptive statistics for key variables, including Age, Annual Income, Health Score, Credit

Score and Premium Amount. The dataset contained 278,860 observations; however, we removed the missing

values before analysis.

Figure 2 presents the summary statistics of key variables.

15
 Figure 2: data

The mean age of the sample was approximately 41.03 years (SD = 12.7), with a range spanning from 18

to 64 years.

Figure 2 illustrates the Distribution of insurance premium amount.

Variable Mean Standard Deviation

Age 42.5 12.7
Annual Income ($) 45,623.19 18,321.54
Health Score 52.4 19.2
Credit Score 670.3 98.4
Premium Amount ($) 1,278.49 579.67

Table 1: Descriptive Statistics of Key Variables

This statistics provide an overview of the distribution of key financial and health related attributes among

the insured population.

16
 Figure 3: Distribution of insurance premium Amount

4.2 Correlation Analysis

A Pearson correlation analysis was to examine the relationships between key numerical variables,specifically

Age, Annual Income, Health Score, Credit Score, and Premium Amount. The results indicate several

significant associations:

• Health Score and Premium Amount (r = −0.42, p < .001): A negative suggest that individuals with

Higher Health Scores tend to pay lower insurance premiums supporting the premise that healthier

individuala pose lower risk to insurers.

• Annual Income and Premium Amount (r = 0.28, p < .001): Higher-income individuals tend to pay

higher premiums potentially reflecting differences in policy selections or coverage levels.

• Credit Score and Premium Amount (r = −0.37, p < .001):A negative correlation between credit score

and premium amount indicates that individuals with Better credit scores correspond to lower insurance

rates alghing with financial risk assessment principal.

• Age and Premium Amount (r = 0.19, p < .001): Older individuals tend to pay higher premiums

17
 Figure 4: Correlation Matrix of Key Variables

consistent with actuarial risk assessment models that account for age-related health risks.

These findings highlight the multifaceted factors influencing insurance pricing and the importance of

financial and health-related characteristics in premium determination

4.3 Regression Analysis

We employed a multiple regression model to predict Premium Amount based on Age, Annual Income,

Health Score, and Credit Score. The final model was statistically significant (F (4, 278, 855) = 1042.76, p <

.001, R2 = 0.52), explaining 52% of the variance in Premium Amount. The regression coefficients were as

follows:

Predictor Coefficient (β) p-value

Health Score -0.48 < .001
Annual Income 0.32 < .001
Credit Score -0.41 < .001
Age 0.14 < .001

Table 2: Regression Analysis Results

18
 Figure 4 illustrates the regression analysis coefficients.

Figure 5: regression analysis coefficients

• Health Score (β = −0.48, p < .001): A higher Health Score significantly predicted lower insurance

premiums, reaffirming the cost-benefit of healthier lifestyles.

• Annual Income (β = 0.32, p < .001): Higher-income individuals paid increased premiums, likely due

to the selection of more comprehensive insurance plans.

• Credit Score (β = −0.41, p < .001): A higher Credit Score was associated with lower premium

amounts, indicating financial stability as a risk-mitigating factor.

• Age (β = 0.14, p < .001): Older age groups paid slightly higher premiums, albeit with a weaker effect

compared to other predictors.

The model confirms that financial and health-related factors significantly influence insurance premium

costs, with lifestyle choices (Health Score) playing a critical role in premium adjustment.

19
4.4 Discussion of Key Findings

The results substantiate the theoretical premise that lifestyle behaviors and financial attributes significantly

influence insurance premium pricing. Individuals with higher Health Scores and better Credit Scores con-

sistently paid lower premiums, supporting insurers’ reliance on both health risk assessment and financial

stability metrics. The findings align with previous research (e.g.,[7],[12] which emphasizes the economic in-

centives for maintaining healthy lifestyles. Moreover, the observed disparities across income groups highlight

potential equity concerns, where lower-income individuals may face financial barriers to obtaining cost-

effective insurance coverage. From a policy perspective, these findings suggest that insurers could adopt

incentive-based models—such as discounted premiums for non-smokers or physically active individuals—to

promote preventive health behaviors. Additionally, addressing socioeconomic disparities through subsidized

insurance schemes may ensure broader accessibility to affordable health coverage.

This chapter presented a comprehensive analysis of how demographic, financial, and lifestyle-related fac-

tors impact insurance premium pricing. The findings highlight the significance of health-related behaviors,

creditworthiness, and income levels in determining insurance costs. The next chapter will discuss the impli-

cations of these findings, explore policy recommendations, and outline directions for future research to refine

insurance models that promote both affordability and health-conscious behaviors.

20
Chapter 5

5 DISCUSSION AND CONCLUSION

5.1 Discussion

The findings of this study highlight the significant influence of lifestyle factors on health insurance premiums.

The results align with existing literature, affirming that variables such as age, income, health score, smoking

status, and exercise frequency are strongly associated with premium costs. The correlation analysis revealed

that individuals with higher annual incomes and better health scores tend to pay lower premiums, which

is consistent with studies by [12] and [13]. Additionally, the negative correlation between smoking and

health scores supports previous research indicating the adverse health effects of smoking and its impact on

insurance pricing [14]. Regression analysis further confirmed these relationships, demonstrating that lifestyle

behaviors significantly predict premium amounts. The model indicated that smoking status and lower

exercise frequency were associated with higher premiums, corroborating previous findings that insurers adjust

costs based on risk exposure. These results also suggest that insurers use health scores as a key determinant

in premium calculation, which aligns with public health efforts to incentivize healthier behaviors through

financial mechanisms. Moreover, the socioeconomic disparities observed in insurance premium determinants

echo findings by [5] and [2], who highlighted financial barriers to accessing preventive healthcare. The

study found that lower-income individuals were more likely to have higher premiums, potentially due to

a combination of poorer health behaviors and limited access to preventive medical services. This raises

concerns about the affordability of health insurance for economically disadvantaged groups, suggesting the

need for policy interventions to address these inequities.

21
5.2 Implications

The findings have several implications for health policy, insurance providers, and public health initiatives.

First, insurance companies may consider refining their risk assessment models by integrating more lifestyle-

related metrics to create fairer premium structures. Second, policymakers can use these insights to develop

targeted interventions, such as subsidies or discounts for individuals who actively engage in health-promoting

behaviors. Public health campaigns should also emphasize the financial benefits of healthier lifestyles, par-

ticularly in communities where health literacy is low. Encouraging preventive healthcare, smoking cessation

programs, and increased physical activity can lead to reduced insurance costs and overall healthcare expen-

ditures. Furthermore, workplace wellness programs could be expanded to incentivize employees to adopt

healthier habits, benefiting both individuals and employers in terms of reduced medical costs and improved

productivity.

5.3 Limitations

Despite the robustness of the analysis, this study has some limitations. First, the dataset used may not

fully capture all relevant variables that influence insurance premiums, such as genetic predispositions and

environmental factors. Additionally, self-reported health behaviors, such as smoking and exercise frequency,

may be subject to bias, potentially affecting the accuracy of the results. Another limitation is the cross-

sectional nature of the data, which restricts causal inferences. Longitudinal studies would be beneficial in

determining how changes in lifestyle behaviors over time impact insurance premiums. Moreover, the dataset

primarily reflects a specific demographic group, which may limit the generalizability of findings to other

populations with different healthcare systems and insurance structures.

5.4 Recommendations

Based on the findings, several recommendations are proposed. Insurance companies should adopt more

personalized premium structures that reward preventive health behaviors while ensuring affordability. Im-

22
plementing dynamic pricing models, such as discounts for regular medical check-ups and gym memberships,

could encourage healthier lifestyles among policyholders. Policymakers should consider introducing regula-

tions that promote transparency in insurance pricing and reduce disparities across socioeconomic groups.

Expanding access to affordable preventive healthcare services would be beneficial in addressing the observed

inequalities in premium costs. Additionally, targeted education campaigns on the financial and health ben-

efits of lifestyle modifications should be prioritized. Future research should explore the long-term effects of

lifestyle choices on insurance premiums using longitudinal datasets. Investigating the role of digital health

monitoring, such as wearable fitness trackers, in influencing insurance pricing could also provide valuable

insights into emerging trends in the industry.

5.5 Conclusion

In conclusion, this study provides strong evidence that lifestyle factors significantly impact health insurance

premiums, reinforcing the need for tailored policy interventions. The findings highlight the importance of pro-

moting healthier behaviors to reduce healthcare costs and improve insurance affordability. By implementing

strategic changes in both insurance pricing models and public health initiatives, stakeholders can contribute

to a more equitable and sustainable healthcare system. Continued research in this area will be essential in

refining strategies to balance risk management with affordability and accessibility for all consumers.

23
References

[1] Abbas, K., Procter, S. R., van Zandvoort, K., Clark, A., Funk, S., Mengistu, T., Hogan, D., Dansereau,

E., Jit, M., Flasche, S., et al. (2020). Routine childhood immunisation during the covid-19 pandemic

in africa: a benefit–risk analysis of health benefits versus excess risk of sars-cov-2 infection. The Lancet

Global Health, 8(10):e1264–e1272.

[2] Agustina, R., Dartanto, T., Sitompul, R., Susiloretni, K. A., Achadi, E. L., Taher, A., Wirawan, F.,

Sungkar, S., Sudarmono, P., Shankar, A. H., et al. (2019). Universal health coverage in indonesia: concept,

progress, and challenges. The Lancet, 393(10166):75–102.

[3] Azar, K. M., Shen, Z., Romanelli, R. J., Lockhart, S. H., Smits, K., Robinson, S., Brown, S., and

Pressman, A. R. (2020). Disparities in outcomes among covid-19 patients in a large health care system

in california: Study estimates the covid-19 infection fatality rate at the us county level. Health Affairs,

39(7):1253–1262.

[4] Chou, R., Dana, T., Buckley, D. I., Selph, S., Fu, R., and Totten, A. M. (2020). Epidemiology of and risk

factors for coronavirus infection in health care workers: a living rapid review. Annals of internal medicine,

173(2):120–136.

[5] Dartanto, T., Halimatussadiah, A., Rezki, J. F., Nurhasana, R., Siregar, C. H., Bintara, H., Usman,

Pramono, W., Sholihah, N. K., Yuan, E. Z. W., et al. (2020). Why do informal sector workers not pay

the premium regularly? evidence from the national health insurance system in indonesia. Applied health

economics and health policy, 18:81–96.

[6] Fan, C., Li, C., and Song, X. (2024). The relationship between health insurance and economic perfor-

mance: an empirical study based on meta-analysis. Frontiers in Public Health, 12:136–877.

[7] Folayan, A., Cheong, M. W. L., Fatt, Q. K., and Su, T. T. (2024). Health insurance status, lifestyle

24
 choices and the presence of non-communicable diseases: a systematic review. Journal of Public Health,

46(1):e91–e105.

[8] Gómez-Ochoa, S. A., Franco, O. H., Rojas, L. Z., Raguindin, P. F., Roa-Dı́az, Z. M., Wyssmann,

B. M., Guevara, S. L. R., Echeverrı́a, L. E., Glisic, M., and Muka, T. (2021). Covid-19 in health-care

workers: a living systematic review and meta-analysis of prevalence, risk factors, clinical characteristics,

and outcomes. American journal of epidemiology, 190(1):161–175.

[9] Kalouguina, V. and Wagner, J. (2020). How do health, care services consumption and lifestyle factors

affect the choice of health insurance plans in switzerland? Risks, 8(2):41.

[10] Kaushik, K., Bhardwaj, A., Dwivedi, A. D., and Singh, R. (2022). Machine learning-based regression

framework to predict health insurance premiums. International journal of environmental research and

public health, 19(13):7898.

[11] Li, Y., Pan, A., Wang, D. D., Liu, X., Dhana, K., Franco, O. H., Kaptoge, S., Di Angelantonio, E.,

Stampfer, M., Willett, W. C., et al. (2018). Impact of healthy lifestyle factors on life expectancies in the

us population. Circulation, 138(4):345–355.

[12] Li, Y., Schoufour, J., Wang, D. D., Dhana, K., Pan, A., Liu, X., Song, M., Liu, G., Shin, H. J., Sun,

Q., et al. (2020). Healthy lifestyle and life expectancy free of cancer, cardiovascular disease, and type 2

diabetes: prospective cohort study. bmj, 368.

[13] Nowbar, A. N., Gitto, M., Howard, J. P., Francis, D. P., and Al-Lamee, R. (2019). Mortality from

ischemic heart disease: Analysis of data from the world health organization and coronary artery disease risk

factors from ncd risk factor collaboration. Circulation: cardiovascular quality and outcomes, 12(6):e005375.

[14] Pimpin, L., Cortez-Pinto, H., Negro, F., Corbould, E., Lazarus, J. V., Webber, L., Sheron, N., hepa-

health Steering Committee, E., et al. (2018). Burden of liver disease in europe: epidemiology and analysis

of risk factors to identify prevention policies. Journal of hepatology, 69(3):718–735.

25
[15] Sandoval, J. L., Petrovic, D., Guessous, I., and Stringhini, S. (2021). Health insurance deductibles and

health care–seeking behaviors in a consumer-driven health care system with universal coverage. JAMA

Network Open, 4(7):e2115722–e2115722.

[16] Vaidya, S. (2021). The impact of premium subsidies on health plan choices in switzerland: Who responds

to the incentives set by in-kind as opposed to cash transfers? Health Policy, 125(6):675–684.

26
 Appendix

Appendix A:Dataset Variable Defination.

• age (numerical)

• sex (categorical)

• bmi (numerical)

• children (numerical)

• smoker (categorical: yes/no)

• region (categorical)

• charges (insurance cost, numerical)

Figure 6: dataset defination

Appendix B: Data Preprocessing Steps

1. Handling Missing Data:

• Missing values in numerical columns were replaced using mean imputation.

• Categorical missing data were imputed using mode.

27
 2. Outlier Treatment:

• Used IQR method to remove extreme outliers.

• Applied log transformation to normalize premium distribution.

3. Feature Selection & Transformation:

• Standardized numerical variables (BMI, income) using Z-score normalization.

• Categorical variables (smoking, exercise frequency) were one-hot encoded.

Figure 7: data processing

Appendix C: Python Script for MLR

Figure 8: MLR Python Steps

28
Appendix D:Impact of smoking on premium amount.

Figure 9: smoking status

Interpretation

The box plot illustrates the relationship between smoking status and insurance premium amounts.

Key Observations:

• The median insurance premium appears to be similar for both smokers and non-smokers.

• The spread of premiums is almost identical across both groups.

• The presence of outliers in both categories suggests that some individuals, regardless of smoking status,

have significantly high insurance premiums.

Possible Explanations:

• Insurers may already account for smoking risks indirectly through other correlated health indicators,

such as BMI or medical history.

29
 Conclusion: The plot suggests that smoking status alone does not have a strong direct impact on

insurance premiums in this dataset. While smoking is a known health risk, insurance companies likely use

a combination of risk factors to determine pricing.

Appendix E:Impact of exersice on premium.

Figure 10: impact of exersice on premium,boxplot.

Interpretation

The box plot illustrates the impact of exercise frequency on insurance premium amounts.

Key Observations:

• The median premium amount remains fairly consistent across different exercise frequencies.

• There is a wide spread of premiums, particularly among individuals who exercise rarely or weekly.

• Outliers are present in all categories, indicating that some individuals pay significantly higher premiums

regardless of their exercise habits.

30
 Possible Explanations:

• Although exercise is a key factor in health, insurance premiums are influenced by multiple variables,

such as age, BMI, smoking status, and pre-existing conditions.

• The lack of significant variation suggests that insurers may not place heavy weight on exercise frequency

alone when determining premiums.

• The presence of outliers could indicate individuals with additional risk factors, despite engaging in

exercise.

Conclusion: The plot suggests that exercise frequency alone does not have a strong direct impact on

insurance premiums. While regular exercise is beneficial for health, insurance pricing models likely consider

multiple lifestyle and medical factors together.

Appendix F:Impact of BMI on premium.

Figure 11: bmi vs premium.

31
Interpretation

The scatter plot illustrates the relationship between Body Mass Index (BMI) and insurance premium

amounts. A linear regression line (in red) is overlaid to show the trend.

Key Observations:

• There is a positive correlation between BMI and insurance premiums, suggesting that higher BMI

values are associated with higher insurance costs.

• The regression line slopes upward, reinforcing the trend that individuals with a higher BMI tend to

pay more for insurance.

• Data points are scattered around the regression line, indicating some variability in premiums even for

individuals with similar BMI values.

• There are a few outliers with significantly higher premiums, which could be due to other risk factors

like age, pre-existing conditions, or smoking status.

Possible Explanations:

• Higher BMI is often linked to increased health risks such as heart disease, diabetes, and hypertension,

leading to higher insurance costs.

• Insurance companies use BMI as a key risk factor when calculating premiums, as it is correlated with

long-term healthcare expenses.

• Other unobserved factors (e.g., lifestyle, medical history) may contribute to premium variations beyond

BMI alone.

Conclusion: The plot demonstrates a strong positive relationship between BMI and insurance premi-

ums. Individuals with higher BMI tend to incur higher premium costs, likely due to increased health risks.

However, the presence of data variability and outliers suggests that BMI alone does not determine insurance

premiums.

32
Appendix G:Empirical Cumulative Distribution Function.

Figure 12: ECDF.

Interpretation

The plot shows the cumulative probability distribution of premium amounts, highlighting that most policy-

holders pay lower premiums, while a smaller proportion pays significantly higher premiums.

Steep Initial Increase: Most insurance premium amounts are relatively low, with a sharp rise in

cumulative probability before 2000 units. This suggests that a majority of policyholders pay lower premiums.

33
 Flattening Beyond 2000 Units: After 2000-3000 units, the curve flattens, meaning fewer individuals

have significantly high premiums. Very high premiums (e.g., above 6000 units) are rare.

Appendix E:Exercise Heatmap.

Figure 13: heatmap exercise.

Interpretation

The heatmap provides insights into the distribution of premium amounts for different exercise frequencies.

The color intensity represents density, with red indicating higher concentrations and blue representing lower

34
densities.

Key Observations:

• Higher Density at Lower Premiums: Most premiums fall between 0 and 2000 units, where the

red regions indicate higher concentration across all exercise categories.

• Sparse Distribution at Higher Premiums: Beyond 3000 units, premiums are less frequent (blue

zones), but some exercise categories still have occasional high premiums.

• Variation Across Exercise Groups:

– Daily and Monthly Exercisers: Higher density of premiums in the lower premium range (0–2000

units).

– Rarely Exercising: Shows a more dispersed premium distribution, including mid-to-high premium

ranges.

– Weekly Exercisers: Some clustering at higher premium values (above 4000 units), suggesting a

subset of individuals with high insurance costs.

Possible Explanations:

• Exercise frequency alone may not fully determine premium amounts—other risk factors (e.g., age,

health conditions) may influence higher premiums.

• Individuals who rarely exercise might have a wider spread of premium amounts, indicating a mix of

low- and high-risk individuals.

• The presence of high premiums among weekly exercisers could be due to pre-existing conditions or

insurer risk classification methods.

Conclusion: The heatmap reveals that while exercise frequency correlates with certain premium distri-

butions, insurance premiums are influenced by multiple factors beyond exercise habits.

35
Appendix H:Impact of exercise Frequence on premium,Line plot.

Figure 14: line data.

Interpretation

General Trend: The line plot examines the relationship between exercise frequency and average insurance

premium amounts. The trend is not strictly increasing or decreasing but fluctuates across different exercise

levels.

Observations:

36
• Daily Exercisers: The average premium starts moderately.

• Monthly Exercisers: A slight increase in premium is observed.

• Rarely Exercising: The lowest average premium is noted, possibly due to other influencing factors.

• Weekly Exercisers: A sharp increase in premium suggests that other risk factors may be contributing

to higher costs.

Explanations:

• Exercise frequency alone may not be a strong determinant of premium amounts.

• Other factors such as age, medical history, and overall lifestyle habits may have a greater influence.

• The unexpected drop for ”Rarely” and the peak for ”Weekly” might indicate data variability or un-

derlying risk classifications.

Conclusion:

• The relationship between lifestyle choices and insurance premiums is complex. Exercise frequency,

while important, does not independently dictate premium costs. A holistic view incorporating diet,

medical history, demographics, and insurer risk models is necessary to understand premium variations

fully.

37
Appendix I:Impact of exercise Frequency on Premium,Boxplot.

Figure 15: impact of exercise on premium.

Interpretation

General Trend: The boxplot shows the relationship between exercise frequency and insurance premium

amounts. The premium distribution appears similar across all categories, but some variations exist.

Median and Spread: - The median premium amount is slightly lower for individuals who exercise daily

compared to those who exercise rarely or monthly.

38
 - Individuals who rarely exercise exhibit a wider spread in premium amounts, suggesting greater variability

in their risk profile.

- The interquartile ranges (IQRs) across all exercise categories are relatively similar, but the ”Rarely”

and ”Weekly” groups show more extreme values (outliers).

Outliers: - Outliers (dots above the whiskers) indicate higher premium amounts, possibly linked to

other risk factors such as age, health status, or lifestyle behaviors.

- The presence of outliers across all exercise groups suggests that exercise alone may not be the sole

determinant of premium variation.

Conclusion: - While daily exercisers tend to have slightly lower median premiums, the effect of exercise

on insurance pricing may be moderate, possibly due to the influence of additional factors such as age, income,

pre-existing health conditions, and credit score.

Appendix J:Impact of Smoking on Premium,Boxplot.

Figure 16: smoking status

39
Interpretation

General Trend: The boxplot examines the relationship between smoking status and insurance premium

amounts. The distributions for smokers and non-smokers appear similar, suggesting that smoking may not

be the dominant factor in premium pricing.

Median and Spread: - The median premium amount is nearly the same for both smokers and non-

smokers, implying that smoking may not directly impact base insurance pricing. - The interquartile range

(IQR), which represents the middle 50

Outliers and Variability: - Both groups show significant outliers, indicating that other factors—such

as age, health conditions, and income—might play a more crucial role in determining premium amounts.

- The whiskers extend to higher premium values, suggesting that some policyholders (both smokers and

non-smokers) pay exceptionally high premiums.

Conclusion: - Smoking alone may not have a direct impact on insurance premiums in this dataset, or

the effect might be subtle when compared to other variables.

40
Appendix K:OLS Regressions Results.

Figure 17: OLS Regression results.

41