We study regression models for nonstationary categorical time series and their applications, and address the issues of prediction, estimation and control. Generalized Linear Models and Partial Likelihood are the basic tools in the present study. The models link the probabilities of each category...