Cluster analysis application in the compulsoryinsurance of civil-legal liability of the vehicles’ owners

Abstract

With an increase in flow of the processed and stored information in insurance organizations in Kazakhstan, associated with the building of customers’ base, mergers and acquisitions processes and implementation of the new insurance products; the relevance of the problem of preliminary information processing for its structuring, allocation of distinctive attributed, generalization and sorting grows. Without appropriate scientific and methodological approach, data processing and analysis will be more difficult for insurance organizations and, may require the utilization of significant informational-computing and financial resources. In the present article as a modern scientific-research approach to the solution of this problem, it is suggested to apply a procedure of the cluster analysis by k-means algorithm, which makes it possible to simplify the processing and further analysis of data set by arranging data in relatively homogeneous groups. Particularly, the present article describes a process of the cluster analysis application by the k-means algorithm to the data on losses by a class of Compulsory insurance of civil-legal liability of the vehicles’ owners. The purpose of the present article is to split the losses by this class of insurance into homogeneous qualitative groups (clusters) based on frequency and severity of losses and, to interpret acquired clusters. Results of the k-means algorithm confirm that each acquired cluster has statistically significant data with similar impact upon losses’ process, which may be employed in the future for evaluation of losses of the insurance organization. Methodological approaches and results obtained in this article will, first of all, be interesting to the professional participants of insurance market of the Republic of Kazakhstan to conduct better underwriting research on the formation of the efficient structure of the insurance portfolio of Compulsory insurance of civil-legal liability of the vehicles’ owners in accordance with tariff rates.

Keywords

• cluster analysis
• unsupervised machine learning
• k-means algorithm
• insurance
• underwriting analysis