We develop a methodology for multivariate time-series analysis when our time-series has components that are both continuous and categorical. Our specific contribution is a logistic smooth-transition regression (LSTR) model, the transition variable of which is related to a categorical time-series...