Ridge Estimator for the Bell-Touchard Regression Model
Ridge Estimator; Bell-Touchard Distribution; Multicollinearity.
The two-parametric discrete distribution family known as the Bell-Touchard distribution has been recently proposed in the statistical literature. It can be used to model count data and stands out for its flexibility, especially in scenarios of overdispersion, where the variance exceeds the mean. The regression model for response variables in count form based on this distribution has also been presented in the literature. The Ridge estimator is widely used to address multicollinearity in regression models, employing an approach that incorporates a penalty on the regression coefficients, thereby reducing prediction errors. In this work, the Ridge estimator will be applied to the Bell-Touchard regression model with the aim of modeling count data in the presence of multicollinearity. Monte Carlo simulations comparing the Ridge estimator with the traditional maximum likelihood estimator will be presented and discussed. Finally, applications to real data will be considered.