不確定性環境下比較灰色預測模型績效

Comparing forecast performance of grey prediction models under the uncertain environment

面對急遽變動的世界，預測的需求有增無減，並且，預測總是被期待有助於達成更佳的規劃與決策，從而降低風險與提升效益。針對不確定性環境下的少數據進行預測分析，灰色預測模型具有獨特的優勢。較常見的GM(1,1)與灰色Verhulst模型之選用，雖然在理論上簡單明確，但在實際應用時，無可避免必須比較灰色預測模型的績效。因此本研究提出一套解決方案，包含預測技術的評選與應用步驟，除強調樣本外的預測績效之外，並自動給出各種灰色預測模型的績效比較結果。因此，本研究並且幫助預測分析者，有系統性而迅速選擇適用的灰色預測模型，從而採取妥善的應對行動。特以陸客來台觀光數據為例，展示本提案的實用性。

Facing up to the sharply changing world, decision makers always consider some kind of forecast that can help them achieve maximizing benefits and minimizing risks for the future. Grey prediction models have unique advantages. in forecasting problems that need to cope with a small data set under the uncertain environment. Although previous works have provided guidance for properly selecting GM (1,1) or grey Verhulst models to use, it is still necessary to pay the efforts of comparing forecast performance between grey prediction models in practice. Therefore, we propose an effective solution which can deal with issues such as selection and performance comparison for grey prediction models, in order to help forecasters and organizations make the best use of grey prediction models in a systematic way. Additionally, a practice study,using the data of Chinese tounist, is presented to illustrate the application of the proposed solution