Amda Report

rising auto insurance claims on the east coast

Advanced Multivariate Data analysis

Project Proposal

Submitted To

Prof. Malay Bhattacharyya

BY – GROUP 9

Paarth Gupta (1211281)

Samidh Pratap (1211296)

Shrey Sahay (1211294)

Indian Institute Of Management Bangalore

Date: 06.12.2013

Introduction

The company’s profits have been decreasing although the revenue has increased. This is because the number of claims has been increasing.

The company wants to figure out what factors affect the likelihood of claim, which segments it should avoid underwriting and which segments are attractive ones to target. The segments can potentially be broken down on the basis of vehicle characteristics, demographic profiles of the policies issued and the behavior of vehicle drivers. We pose the following problems:

1. Mention three parameters which most likely affect the likelihood of claim. This would be the starting point before any other kind of analysis that can be done. Most of the data uses categorical variables to describe the car, geography and driver characteristics.

2. Suggest segments which the company should avoid underwriting in the future. In effect, this is the objective of the company since it is looking to reduce claims and target appropriate segments.

3. Identify the low risk segments. These segments would represent the future strategy of the company.

Data set

The data set gives the details of an insurance company which provides car insurance in seven states along the east coast of USA. The dataset gives the details of the driver, vehicle and the state the car is registered in. The data set has been procured from EXLService Holdings, Inc and suitable modifications have been made to the data for the purpose of the analysis. The name of the company in the analysis has been withheld for privacy reasons. Other techniques may be used during the analysis based on the need.

The various fields are as described below:

Name | Definition |...