On Multi-dimensional Markov Chain Models

Abstract

Markov chain models are commonly used to model categorical data sequences. In the paper, we propose a multi-dimensional Markov chain model for modeling high dimensional categorical data sequences. In particular, the model is practical when there are limited data available. We then test the model with some practical sales demand data. Numerical results indicate the proposed model when compared to the existing models has comparable performance but has much less number of model parameters.