SIMULATION OF COMPARATIVE STUDY OF JAMES-STEIN ESTIMATOR, RIDGE REGRESSION ESTIMATOR, AND MODIFIED KIBRIA LUKMAN ESTIMATOR IN HANDLING MULTICOLLINEARITY IN POISSON REGRESSION

Poisson regression is a statistical method used to analyze data with a response in the form of a count variable. The purpose of this study is to compare the performance of the Poisson James-Stein Estimator, Poisson Ridge Regression Estimator, and Poisson Modified Kibria-Lukman Estimator methods in dealing with multicollinearity using simulated data with n = 20, 40, 60 and 80 in poisson model (p=6) with 𝜌 = 0.3 and 0.99. The best model was compered based on the MSE value. The results showed that in the partial correlation, PRRE method of k2 parameters better in overcoming multicollinearity at n = 20 and PMKLE parameters k2 better in overcoming multicollinearity at n = 40, 60, and 80 and in the full correlation data, PRRE method of k2 parameters better in overcoming multicollinearity at n = 20 and 40, and PMKLE parameters k2 were better in overcoming multicollinearity at n = 60 and 80.