Document Type : Original Research Paper
Authors
Department of Information Technology Management, Faculty of Computer Engineering and Information Technology, Payame Noor University of Tehran, Tehran, Iran
Abstract
The third person car insurance has the biggest portion in the insurance market which makes an appropriate opportunity for data discovery and extracting unknown patterns for decision making in insurance industry. Currently premium is calculated with the minimum consideration to the risks. Based on these calculations lots of damages may occur to the insurers which affects the quality of their services which also leads to costumers’ dissatisfaction. To make data mining happen, vehicles’ information, police background and the insured person’s information are gathered, standardized and stored in a data warehouse from over 30 million policies and 2.7 million losses. For standardizing vehicles’ specification, police data are used and insured people’s identity are evaluated with national databases. Thereafter a mining structure designed and the three algorithms of clustering, neural network, and decision tree were performed on it. Finally, all models are evaluated using sample data and the results checked with confusion matrix and loss rate, which indicates the feasibility of this method in dynamically tariffing for this type of policies and leads to the decrease of the loss rate, the confusion matrix also indicates the accuracy of the evaluation.
Keywords
Letters to Editor
Send comment about this article