應用一般迴歸類神經網路與序列探勘技術建構企業財務危機預警模型－以台灣電子產業為例

Applying General Regression Neural Network and Sequential Pattern Mining to Construct Corporation Financial Crisis Predictive Model－A Case of Taiwan’s Electronic Industry

蔡明賢(Ming-Hsien Tsai)

指導教授 ： 羅淑娟

國立臺北科技大學/管理學院/工業工程與管理研究所/碩士(2008年)

https://doi.org/10.6841/NTUT.2008.00443

摘要

由於台灣產業結構變化，電子產業在政府政策發展下，已成為台灣產業中眾所矚目的明星，因而使得股票市場上的資金紛紛往電子類股靠攏，然而電子產業景氣變動循環迅速，一旦企業發生財務危機，影響所及，不僅會造成股市動盪，就連一般投資大眾、金融機構、往來廠商及員工生計也將受其拖累，更會影響整體金融體系的運作，成為整體台灣經濟之一大隱憂。而目前的財務危機分類研究中，其分類結果是依機率決定所屬類別，但公司的財務惡化現象具有時間序列關係，若分類模式能納入資料的序列性資訊，應可彌補分類模式在預測能力上的不足。
本研究試著以一般迴歸類神經網路整合序列探勘技術，利用序列探勘技術對類神經網路的預測值進行探勘進而建立一套較簡易、有效且針對台灣電子產業的企業早期危機預警模型，並由實驗顯示出一般迴歸類神經網路結合序列探勘技術之整合模型，在缺乏現有資料之情形下，的確能提供未來財務狀況一個有前瞻性預測的參考價值。

關鍵字

一般迴歸類神經網路 ； 序列樣式探勘 ； 財務危機預警

並列摘要

Due to the variety of industrial structure, electronic industry which is facilitated by government strategies has become a remarkable industry in Taiwan. And this made many people invest their money in electronic industry in the stock market. Since the economic circulation of electronic industry varies rapidly, it will bring a huge impact on the stock market if financial crises of enterprise happen once. Many investors, financial institutions, related companies, and employees will suffer from this situation. Even all the operations of monetary system in Taiwan will also affected by the financial crisis of electronic industry, which will be a great worry to entire economy of Taiwan.
In recent classification researches of financial crisis, the result has depended on probability. In fact the financial crisis deterioration relates to time series. We could help the inefficiency of predictive ability of classification model if classification model can integrate financial time series data. This study will integrate general regression neural network and sequential pattern mining technique to construct an early warning predictive model, which takes advantage of sequential pattern mining technique to extract predictive sequential pattern from predictive value of GRNN. The experimental result showed that the integrated model really can predict an effective and reliable result of financial situation in the future without present data.

並列關鍵字

GRNN ； Sequential Pattern Mining ； Financial Crisis Prediction

參考文獻

[2] 陳仕昇、陳彥良、許秉瑜，「以可重複序列探勘網路瀏覽規則之研究」，資管評論，第九期，1999。

[4] R. Agrawal and R. Srikant, “Fast Algorithm for Mining Association Rules,” in Proc.20th VLDB conference ,Santiago, Chile, 1994.

[5] E.I. Altman, “Financial ratios, Discrimnant analysis and the prediction of corporate bankruptcy,” The Journal of Finance, vol. 23, no. 4, 1968, pp. 589-609.

[6] W.H. Beaver, “Financial Ratios as Predictors of Failure,” Journal of Accounting Research, Vol. 4, 1966, pp. 71-111.

[7] G.J. Bowden, J.B. Nixon, G.C. Dandy, H.R. Maier and M. Holmes, “Forecasting chlorine residuals in a water distribution system using a general regression neural network,” Mathematical and Computer Modelling, vol 44, 2006, pp. 469–484.