Overview

This project develops a motor insurance pricing model using actuarial techniques. The objective is to estimate the expected cost of claims for motor insurance policies by modelling both claim frequency and claim severity.

The analysis is based on a Kenyan motor insurance dataset and applies Generalized Linear Models (GLMs), a standard approach used in actuarial pricing.

Methodology

The pricing framework follows the classical actuarial approach:

Expected Claim Cost = Frequency × Severity

1. Frequency Model

• Modelled using a Poisson GLM
• Estimates the expected number of claims per policy
• Key predictors include vehicle type and engine capacity

2. Severity Model

• Modelled using a Gamma GLM
• Estimates the average cost of claims
• Appropriate for skewed insurance loss data

3. Pure Premium

• Calculated as:

  Pure Premium = Predicted Frequency × Predicted Severity

• Represents the expected claim cost per policy

Key Insights

• Motorcycles exhibit higher expected claim costs compared to private vehicles
• Vehicle characteristics significantly influence both frequency and severity