灰色預測能對於少數據與不確定資料做有效的處理,並具有不錯的準確性。目前已有許多學者研發出不同的灰色預測模型中,但還是延伸出兩類新的研究問題。針對第一類的問題,目前並未發現使用灰色預測模式於區間值資料預測方面的研究。至於第二類的問題,目前僅有幾位學者發展的灰色預測模式能提供區間預測值。 針對這兩類問題,本研究擬在少數據不確定時間數列資料下,發展一套完整的灰色區間預測模型,主要是整合區間灰數運算、傳統預測模式、分群方法(線性迴歸法)、GM(1, 1)模式與NGBM(1,1)模式,提出六種灰色區間預測模式。其中四種模式是針對第一類的問題,而另外二種模式則是針對第二類的問題。這六種灰色區間預測模式將能提供決策者精確的預測區間範圍,使決策者能做出更正確的決定。 此外,本研究將以多種不同的少數據區間值與單點值資料,分別測試本計畫預測方法與先前學者的預測方法,比較其預測準確度並評估其優劣。最後,本研究將以兩個實際案例分別驗證本研究預測方法的準確性與實用性。
Grey forecast can effectively deal with small size and uncertain data and provide precise forecast values. Currently, numerous researchers have developed various grey forecast models. Many grey forecasting models have been developed by scholars but two new research issues have emerged. For the former problem, there are rare grey forecast studies for forecasting the interval-value data. For the latter problem, only several grey forecast studies can yield interval-value forecasted results. To aim at solving the two problems, this study attempts to develop an integrated grey interval-value forecast method for small size and uncertain time series data. This method combines grey interval numbers, traditional forecast methods, clustering methods (linear regression method), the GM (1, 1) model and the NGBM (1, 1) model to develop six grey interval forecast methods. The first four methods are designed for the former problem, and the last two is designed for the latter problem. The six grey forecast methods can provide decision analysts with precise forecast range which allows them to make a right decision. In addition, several small size interval-value and single-value data will be employed to evaluate the forecast accuracy between the proposed method and the previous grey forecast methods. Finally, two real-world cases are adopted for validating the forecast accuracy and practicability of the proposed method.