CLAIM DATA ANALYSIS WITH GENERALIZED LINEAR MODEL

• Main Author: LUH PUTU ASRI CAHYANI (NIM: 10113061), NI
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/23495
• Institution: Institut Teknologi Bandung

In the insurance business, claim data have observational values that tend to be large. Therefore, to see the behavior of the insurance claims data required appropriate methods and take into account other variables which can affect the claim amount. In addition, of course, the world of insurance business needs to anticipate the claims in the future. So we need a method to predict that which in this study used Generalized Linear Model. The predictive determination of the claims in this case takes into account some predictor variables that exist in personal injury claims such as accident rate, presence or absence of legal representation, and operational time. Prediction of claim amounts will be useful as a claim backup as an anticipatory action if there is a claim event in the future. There are currently known methods for building a claim reserve. One of the most popular is the Chain-Ladder method. However, the method has not taken into account the distribution of claim amounts. In this study another alternative method will be used that is by using Generalized Linear Model. Then, method of reserving claim using Chain-Ladder method and Generalized Linear Model will be compared. The comparison showed that the raw data which directly built in the form of a run-off triangle and then using a generalized linear model for the claim reserve will generate an error and the mean square error with the smallest value compared to the other three forms of claims reserves.