Granger Causality Networks for Categorical Time Series

We present two model-based methods for learning Granger causality networks for multivariate categorical time series. Our first proposal is based on the mixture transition distribution (MTD) model.