Evaluation of generalized linear models for modelling claim frequencies in vehicle insurance
Date
2020
Authors
Kinyanjui_, Wanjiru;
Journal Title
Journal ISSN
Volume Title
Publisher
Strathmore University
Abstract
This study aims at evaluating the generalised linear models used for
modelling vehicle insurance claim frequency. Modelling claim frequency, in
turn, helps in the pricing and estimation of premiums. In this paper, claim
frequencies will be modelled with respect to other risk factors present in the
vehicle insurance data. This study makes use of data present in the
CASdatasets that can be downloaded as an R package. The specific data used
is brvehins1e that contains 393,071 observations. I further went ahead to
randomly select 10,000 observations for computational purposes.
The four generalised linear models namely; Poisson, negative Binomial, zero
inflated Poisson and zero inflated negative Binomial models were fitted to
the data to evaluate how well they fit. Comparison of the models was done
using the Akaike' s Information Criteria, Bayesian Schwartz Information
_Criteria and as well as performing a Vuong Test. Significant variables ii1. the
model were determined using the p-values. The negative binomial model
was determined as the better model when compared to the Poisson model.
The zero inflated negative Binomial model was also seen to provide a better
fit compared to the zero inflated Poisson model.
Description
Submitted in partial fulfilment of the requirements for the Degree of Bachelor of Business Science in Actuarial Science at Strathmore University