多變量模糊時間數列模式之應用：以臺灣地區國中教師人數之預測為例

Application of Multivariable Fuzzy Time Series Models: Forecasting the Number of Junior High School Teachers in Taiwan Area

本文提出以多變量模糊時間數列模式--二因子模式、引導式模式、馬可夫模式，建構臺灣地區國中教師人數之預測模式，並與類神經網路比較預測結果。實證結果得知在有限資料筆數下，多變量模糊時間數列二因子模式預測結果較佳，類神經網路次之，多變量模糊時間數列馬可夫模式第三。而在方法操作上，以多變量模糊時間數列引導式模式最為簡易。

In this paper， we use multivariable fuzzy time series models and Neural Networks to forecast the number of teachers in junior high school of Taiwan area. The empirical result of analyzing here can be concluded as follows: The error of multivariable fuzzy time series Two-factor model is the smallest， Neural Networks is the second smallest， and multivariable fuzzy time series Markov model is the third smallest. The multivariable fuzzy time series Heuristic model is the easiest method to follow.