Data Warehousing 及 Data Mining 在營建財務及成本管理之決策支援之研究(I)

Abstract

資料採擷(Data Mining)於1990 年開始蓬勃發展之際，1992 年M.Y.便首度以ID3 之演 算法應用於財務危機預測上，其以事實導向方式選取變數，除可更全方位的考量影響變數 外，更可避免專家經驗選取變數之人為偏差，本研究首先將就財務之資料，驗證資料採擷 (Data Mining)進行之分析結果，優於過去以專家經驗選擇較佳變數後再進行分析之結果。 未來一年，本研究期以John Argenti (1976)所提出之『企業生存因素』(Factors Affecting Survival of Business)作為研究背景理論，期能在上百個與營建業相關之變數中以序列分 析(Sequence Pattern Analysis)提出財務危機發生之階段序列型樣(Sequence Pattern)， 再採用集群(Cluster)與ID3 決策樹(Decision Tree)，選取各階段財務危機預測之相關變 數與其關聯性，以作為營建業者自我省察財務狀況之指標，更可作為資金提供者之信用評 估指標，進而降低財務危機之發生，鞏固營建業與金融體系財務之健全。而成本管理之相 關研究亦於本研究成果之後續發展中說明研究進度。

The purpose of this study is to develop a dynamic prediction model for financial distress in construction industry. Three main issues are discussed in this study. First, we study the basic process of financial distress in construction and identify the dependencies used for sequence pattern technique. Second, this study compares two methodologies in financial distress prediction for their accuracy and stability. Lastly, the performances of various distress prediction approaches will be evaluated and compared. The impacts of economic, industrial, and business characteristics on prediction performance will be assessed. Real world financial distress data and statistics from the construction companies in Taiwan are used to formulate and validate the distress prediction models derived in this study. This research expects to provide a quantitative framework in evaluating the financial standing of firms in construction industry. The financial standing evaluation information is crucial to both firm’s managers for understanding the clients and suppliers, and fund suppliers for assessing the credit worthiness of the firm. The early awareness of a potential financial distress can also help to reduce the overall numbers of actual distress in construction industry.