New ridge parameter estimators for the zero-inflated Conway Maxwell Poisson ridge regression model

Abstract

One of the flexible count data models for dealing with over and under-dispersion with extra zeroes is the zero-inflated Conway–Maxwell Poisson (ZICOMP). The ZICOMP regression coefficients are generally estimated using the maximum likelihood estimator (MLE). In the ZICOMP regression model, when the explanatory variables are correlated, the MLE does not give efficient results. To overcome the effect of multicollinearitymode in the ZICOPM regression, we proposed the ridge regression estimator. To evaluate the performance of the estimator, we use mean squared error (MSE) as the performance evaluation criteria. A theoretical comparison of the ridge estimator with MLE is made to show the superiority of the estimator. The proposed estimator is evaluated with the help of a simulation study and a real application. The results of the simulation study and real application show the superiority of the proposed estimator because it produces a smaller MSE as compared to the MLE.

Keywords:

• Conway-Maxwell Poisson regression
• count data
• dispersion
• maximum likelihood estimator
• multicollinearity
• ridge estimator
• zero inflation

Disclosure statement

No potential conflict of interest was reported by the author(s).