Fuzzy Time Series Model Incorporating Predictor Variables and Interval Partition

Prediction is a critical component in decision-making process for business management. Fuzzy Markov model is a common approach for dealing with the prediction of time series. However, not many studies devoted their attention to the effect of the parameters on model fitting for fuzzy Markov model. In the paper, we examine the prediction ability for fuzzy Markov model, based on the data of Taiwan’s exports and foreign exchange rate. The empirical results indicate that fuzzy Markov model performs better for longer period forecasting; moreover, neither increment information nor increasing window basis would improve the performance for fuzzy Markov model. An advantage of the paper provides a beneficial knowledge when using Markov model for prediction. Key-Words: Fuzzy time series, Fuzzy Markov model, High order fuzzy relationship, Increment information, Interval partition, Taiwan exports.