COMPARISON OF MOMENT AND MAXIMUM LIKELIHOOD METHODS IN ESTIMATING PARAMETER GAMMA DISTRIBUTION

Abstract

Claim prediction plays a crucial role in the insurance industry, enabling companies to design suitable insurance policies for potential policyholders. One approach to predicting claims involves using the parameters of the gamma distribution. Several methods can be applied, including Maximum Likelihood Estimation (MLE), the Method of Moments (MoM), and the Bayesian method. This study focuses on comparing the MoM and MLE methods to determine the most effective approach for predicting insurance claim frequency using Google Collab. The analysis is based on secondary data obtained from Kaggle. The MoM estimates parameters by equating k sample moments with the corresponding k population moments, while MLE works by maximizing the likelihood function. The findings indicate that MLE produces a lower error rate of 0.424%, compared to 0.6845% for MoM. This suggests that Maximum Likelihood Estimation (MLE) provides higher accuracy in predicting insurance claim frequency.