本研究分析國內研究所碩士班學生現況，並以時間數列之方法預測未來10年國內研究所碩士班人數，以供各學術單位與各界市場參考，並作為往後招生數量之參考。研究方法以「時間數列推估方法」之ARIMA（Autogressive Integrated MovingAverage，自我迴歸整合移動平均）進行。進行預測之前，先進行模式辨識，以一次差分（first-order differencing）進行模式辨識，以「自我相關函數(q值)」與「偏自我相關函數(p值)」達到顯著的期數來判斷自我相關與移動平均的「秩」（order）。以此p、q值進行模式適合度檢定，若殘差的自我相關與偏自我相關皆未達到顯著，則可代表模式獲得適合。研究結果顯示國內研究所碩士班人數趨勢預測在p=2、d=1、q=3時達到模式穩定並達到適合，並預測出未來10年的碩士班人數。最後根據研究結果，提出結論與建議。

This research analyses the present situation of domestic Master’s students and aims to predict the number of enrolled students in the following 10 years via utilizing Time Series Analysis.This study employs ARIMA (Autogressive Integrated Moving Average), a model of time series analysis, to predict the number of students. Prior to making a prediction,
the model was identified with first-order differencing. The significant number of Autocorrelation Function (q value) and Partial Autocorrelation Function (p value) was used to investigate the order of the autocorrelation and the moving average and to examine the fitness of the model. If the residual of autocorrelation and partial autocorrelation do not reach significance, the model would be appropriate. The results
indicate that the stability of the model was reached when the trend lied in p=2, d=1, q=3,which predicted the number of Master’s students in the following ten years.