Robust modified jackknife ridge estimator for the Poisson regression model with multicollinearity and outliers
Kingsley Chinedu Arum, Fidelis Ifeanyi Ugwuowo, Henrietta Ebele Oranye
Abstract
The parameters in the Poisson regression model are usually estimated using the maximum likelihood estimator (MLE). MLE suffers a breakdown when there is either multicollinearity or outliers in the Poisson regression model. The ridge estimator and the modified jackknife ridge estimator are generally preferred to handle multicollinearity. However, both methods suffer breakdown when there are outliers in the y-direction. Thus, we combined the modified jackknife ridge estimator with the transformed M-estimator (MT) to account for both problems. We carried out a simulation and application study to investigate its performance. The simulation and application results show that the proposed dominates the existing methods (MLE, Ridge, and Modified jackknife ridge).