對於許多傳統的餐飲業者來說，精準的計算出未來欲訂貨的數量多寡能夠有效的控管相關的營運成本。因此，自動化訂貨數量預測是目前許多餐飲業者所要導入的人工智慧技術之一。本研究將針對個案公司的需求進行開發客製化訂貨預測系統，而研究的重點在於從現有的機器學習技術當中找出最佳的預測演算法(包含單一、多重與混合技術)，同時系統將可依照所經營的聯鎖店其地點、店型與時間三種構面進行客製化的訂貨數量預測以達到更有效的決策支援。 ;In the catering industry, it is important to accurately predict the order amount in order to effectively control and manage operating costs. Therefore, many companies have implemented automatic order prediction as one of the artificial intelligence techniques. In this project, a customized order prediction system will be developed for the case company. Particularly, the research focus is on identifying the best prediction techniques from existing machine learning algorithms, including single, ensemble, and hybrid techniques. Moreover, the system will be designed for the specific prediction of order amounts in terms of the locations, types, and time of chain stores, which can reach the more effective decision making purpose.
