Risk Analysis and Accident Pricing with Glm and Aft of Autonomous Cars

摘要

In non-life insurance,claim frequencies are acceptably modeled with generalized linear models(GLMs).This supports in generalizing on the assumption that all policies have a homogeneous risk given that they have the same covariates.The commercializing of autonomous cars falls under this same category of homogeneous risk but the parameters under observation are significantly different from traditional cars(cars with a driver).Since the non-life insurance landscape is transforming,it has become necessary for auto insurers to adopt quick and nimble attention to tackle this disruptive force.For any insurer to be ahead of the race,it is essential to adopt a flexible pricing model to cater for the possible shift in the insurance ecosystem.Base on this problem,this paper develops a framework for working in the direction of estimating the premium and discussing the risk of autonomous cars.In this paper,two models are adopted to give an estimate of the premium value.The first is the accelerated failure model(AFT)under the Weibull distribution:to model for the time taken for the first accident to be recorded.The generalized linear model(GLM)was also used to control for the severity of each accident occurring.Finally,the model is structured by combining the AFT and the GLM.It is revealed in our preliminary analysis that;Lexus autonomous cars prove to be more durable than the other cars.Also,it was better to test cars in Santa Clara than San Francisco.This thesis also contains a brief discussion on the possible risk associated with autonomous cars and a proposed insurance model.

目录

声明

ACKNOWLEDGMENT

ABSTRACT

TABLE OF CONTENT

KEYWORDS

LIST OF ABBREVIATIONS

CHAPTER ONE：INTRODUCTION

Research Question

Objectives

Structure of the thesis

Creativity and Contribution to This Thesis

CHAPTER 2：LITERATURE REVIEW

Autonomous car

Levels of Autonomous cars

Impact on Insurance

Timeline of the Inception of Autonomous cars

Possible Risks Associated with Autonomous cars and the Insurance Industry

Autonomous car risk

Insurance Company Risk

Autonomous car Accident

Liability Analysis of Self-Driven Cars

Generalized Linear Model

Accelerated Failure Time

Commonly Used AFT Models

CHAPTER 3：ACTUARIAL PROBABILITY THEORIES

Generalized Linear Model

Exponential Family

Accelerated Failure Time Model(AFT)-Weibull Distribution

CHAPTER 4：MODELING AND DATA ANALYSIS

Descriptive Analysis

AFT Analysis

GLM Analysis

AFT and GML Output

CHAPTER 5：CONCLUSION AND RECOMMENDATIONS

Limitation

Acknowledgments

REFERENCES