流通業協同規劃預測補貨解決方案

Abstract

在現今競爭激烈的商業環境中，企業必須針對供應鏈進行整體的規劃，並整合供應鏈內成員彼此的政策與行為，方能達到整體供應鏈的利益最大化。有鑑於此趨勢，自願性跨產業商務標準協會於2002年提出協同規劃預測補貨之機制，欲以網頁基底的公開資訊，協調供應鏈成員之間眾多的活動，有效提升整體供應鏈之效益。而本研究即是針對流通業之協同規劃預測補貨，提出有效之解決方案。 由於協同規劃預測補貨屬於概要的商業合作流程，沒有針對產業特性進行修正，亦無詳細定義關鍵流程之執行方式，故企業在實際執行時常遭遇問題。此外，由於關鍵步驟之執行過於複雜，目前尚未有相關研究提出有效輔助之解決方案。因此，本研究針對流通業的供應鏈協同合作架構，分析並修正協同規劃預測補貨之標準概要流程，並加入資訊回饋之機制，提出一個符合流通業需求並能實際執行之協同規劃預測補貨流程，並針對該流程中關鍵且複雜的步驟提出對應之解決方案。 本研究提出之流通業協同規劃預測補貨流程包含七個主要步驟：前端合作協議、產生聯合商業計畫、建立中長期需求預測共識、短期需求預測共識、訂單預測共識、產生與執行訂單，以及成果追蹤分析與修正。其中各步驟皆會依不同的供應鏈架構類型而有所修正，步驟之間亦加入資訊回饋與修正的動作，以補強現有標準流程之不足。 本研究提出之解決方案主要包含三部分：銷售預測、訂補貨建議與績效評估。銷售預測階段首先是資料的分析與整理，而後採用自動學習之方式建立各類商品的最佳預測模型，再以最佳模型進行未來銷售之預測。訂補貨建議階段會考慮商品政策與產品結構樹等資訊，將銷售預測值先轉換為實際的訂補貨需求，再轉換為物料規劃需求，並加總為規劃採購單。績效評估階段則是利用關鍵績效指標之計算，評估協同合作之成效。 最後，本研究建立一協同規劃預測補貨解決方案系統，進行情境分析與實際案例測試。情境分析之結果顯示，本研究之演算法的表現明顯優於一般企業所採用的預測方法；而實際針對台灣流通產業進行導入測試之結果，在各項關鍵績效指標均有相當良好的表現，證明本研究之流通業協同規劃預測補貨流程與相關演算法，實為一可行且有效之解決方案。

This study examines Collaborative Planning Forecasting and Replenishment (CPFR), a web-based tool proposed by Voluntary Inter-Industry Commerce Standards (VICS) in 2002 to coordinate the various supply chain management activities. A varied number of steps in the CPFR process have been defined depending upon the level of detail and CPFR uses a cyclic and iterative approach to derive consensus future forecasts. Nevertheless, many obstacles to CPFR implementation have been reported in the literature because supply chain partners are skeptical and resistant to change. Therefore, this study aims at proposing a practicable and suitable CPFR process for Taiwan’s retail supply chain through analyzing and modifying the various CPFR processes defined previously. Furthermore, because the main purpose of CPFR is to provide the consensus future forecast, this study suggests an effective CPFR solution to help generating this information in the crucial and complex steps of the proposed CPFR process. The CPFR process for Taiwan’s retail supply chain consists of seven steps: developing collaborative agreement; creating joint business plan; creating consistent long-term sales forecast; creating consistent short-term sales forecast; creating consistent order forecast; confirming and generating order; and performance analysis. This study defines each step differently in detail for different supply chain structures and adds information feedback and reversed mechanisms between steps to enhance the original CPFR process. The CPFR solution proposed in this study consists of three modules: (1) Sales Forecasting, including input data analysis, automatic best forecast method learning, and future sales forecast calculation; (2) Replenishment Suggestion, including replenishment calculation and planned purchase order generation; (3) Performance Evaluation, including key performance index (KPI) calculation to evaluate the results of the collaboration. To show the effectiveness and efficiency of the proposed CPFR process and solution, a prototype was constructed and tested to demonstrate the power of the proposed CPFR process and solution, using scenario analysis and three real cases. As results, this CPFR process and solution for Taiwan’s retail supply chain can be shown to have excellent KPI performance in each scenario and each real case.