臺灣博碩士論文加值系統

預測活動對於人類的生活而言是相當重要的，例如天氣預測，地震預測
等。由於傳統的預測方法不能處理歷史資料為語意值的預測問題，利用乏
晰時間系列理論可以來彌補傳統方法之不足。在本論文中，我們提出了兩
個乏晰時間系列模式來處理預測問題，它們分別為非時變一個因素的乏晰
時間系列模式和非時變兩個因素的乏晰時間系列模式。根據所提出的非時
變一個因素的乏晰時間系列模式，我們提出兩個非時變一個因素的乏晰時
間系列演算法(演算法-A和演算法-A*)以預測美國阿拉巴馬州立大學的學
生註冊人數。根據所提出的非時變兩個因素的乏晰時間系列模式，我們提
出兩個非時變兩個因素的乏晰時間系列演算法(演算法-B和演算法-B*)以
預測氣溫。在本論文中所提的演算法不僅具有計算複雜度低的優點，同時
也能得到很好的預測結果。

Forecasting activities are very important to our life, such as
the weather forecasting, the earthquake forecasting, …, etc.
The disadvantage of the traditional forecasting methods is that
it can not deal with forecasting problems in which the
historical data are represented by linguistic values. Using
fuzzy time series to deal with the forecasting problems can
overcome this drawback. In this thesis, we propose two new fuzzy
time series models to deal with forecasting problems, which are
the one-factor time-variant fuzzy time series model and the two-
factors time-variant fuzzy time series model. Based on the
proposed one-factor time-variant fuzzy time series model, we
propose two one-factor time-variant fuzzy time series algorithms
(i.e Algorithm-A and Algorithm-A*) to forecast the enrollments
of the University of Alabama. Based on the proposed two-factors
time-variant fuzzy time series model, we propose two two-factors
time-variant fuzzy time series algorithms (i.e., Algorithm-B and
Algorithm-B*) for temperature prediction. Both of these
algorithms have the advantages of low time complexity and can
get good forecasting results.