運用關聯規則探勘輔助生產管理決策-以案例公司為例

Abstract

本研究以文獻探討與個案資料彙整臺灣TFT-LCD產業內外在經營環境建立產業領域知識，並以生產管理觀點分析生產管理人員在日常生產管理決策面臨資料繁雜與時間有限下，如何下達有效決策以妥善資源調配。

依個案需求分析提出一個運用關聯規則資料探勘輔助生產管理決策系統雛型設計，期望透過跨產業資料探勘標準作業流程來建立資料擷取、處理、分析、模型修正、等流程。運用關聯規則演算法來處理短期生產管理作業在製品生產狀況分析，同時說明如何使用R軟體來建構資料分析系統架構雛型，經由資料分析結果來輔助生產管理人員進行生產資源調配及問題發現與推動改善行動。

最後以資訊經濟觀點透過VPRC策略模型，分析資料探勘系統為生產管理決策帶來那些價值創造與效率改善，以實驗案例來說明資料探勘演算法與領域知識結合為企業帶來的商業價值。

In this study, we collect literature and case study data and outline the operating environment TFT-LCD industry in Taiwan for the establishment of industrial domain knowledge. From the production management point of view, we analyze how production management staffs in charge of the daily production management decisions do effective decision-making in arranging allocation of resources, facing the constraints of data complexity and limited available time. Grounded on the case analysis, we propose an association rules data mining aided production management decision-making system prototype for data acquisition, processing, analyzing, model updating, and other processes through cross-industry standard data mining processes (CRISP-DM). We use association rules algorithms to deal with short-term production management and WIP situation analysis, and explain how to use R (data mining software) to construct data analysis software system architecture prototype, which utilizes data analysis to assist production management managers for decision-making and resource allocation problems discovery and promotion corrective action. Finally, from the point of view of information economy, we apply the VPRC strategy model to analyze how data mining systems can create values and improve efficiency for the production management decision. We use experiments to show how data mining algorithms combined with domain knowledge can bring business value to the enterprise.