TWO PARAMETER RIDGE ESTIMATOR FOR THE BELL REGRESSION MODEL

One solution to the multicollinearity problem in the Bell regression model, which is utilized for over-dispersion issues, is biased estimators. In recent years, some biased estimators have been proposed in the Bell regression model that can be used in modelling correlated count data. In this article, Bell two-parameter ridge estimator (BTPRE) is proposed. This two-parameter estimator has some advantages over the previously proposed estimators. More efficient results are obtained than the Maximum Likelihood estimator (MLE) and Bell Ridge estimator (BRE) in the case of multicollinearity by using BT- PRE. Monte Carlo simulation study and real data results are obtained to show that the proposed estimator is better. Estimators have been compared according to the Mean Squared Error (MSE) criterion. BTPRE is superior to other estimators.