Classification of the insurance sector with logistic regression

Abstract

In cases when categorical results such as successful-unsuccessftil, ill-not ill, good-fair-bad are obtained especially as a result of evaluation of data, the logistic regression is a rather suitable statistical method. In this study, the data of 53 companies that are active in the insurance sector in Turkey for 2004, 2005 and 2006 years were evaluated by using logistic regression method. However, as the data were not sufficient for all the insurance companies, 12 insurance companies were eliminated from the evaluation. 41 companies used for the analysis were divided into two groups depending on their activity area .17 companies were evaluated by using the data on individual accident, health and life branches and 24 companies by using data on fire, transportation, engineering, agriculture, all-risks, obligatory traffic, obligatory highway transportation, individual accident and other accident and health branches.