信用評等制度在金融市場上已近成熟，其所提供之資訊有利於投資人或是金融機構降低因企業產生違約時所受到之損失，為避免發生如金融海嘯時，有較高信用等級之企業，卻發生倒閉，使投資人遭受龐大損失，因此，提供有效之信用評等模型為重要議題。由於我國股市交易量中電子產業占有較高比重，故本研究樣本蒐集2006年至2014年間之電子產業作為研究對象，結合資料探勘技術中之類神經網路、支援向量機、決策樹CART及決策樹CHAID建立分類模型。實證結果顯示，以逐步迴歸篩選搭配類神經網路建構之模型準確率最高，達85.39%。
Credit rating system is nearly mature in the financial markets, the information they provide is conducive to investors or financial institutions to reduce losses due to default when enterprises have suffered, in order to refrain such financial crisis, having a higher credit rating of the enterprise, but the collapse occurred, so that investors have suffered huge losses, and therefore providing valid credit rating model is an important issue. Because of Taiwan's stock market trading volume in the electronics industry occupies a higher proportion. Therefore, this research sample are collected with electronics industry from 2006 to 2014 as a research object. Combining with data mining techniques like Neural Networks, Support Vector Machines, Classification and Regression (CART), and Chi-square Automatic Interaction Detection (CHAID) to construct a classification model. According to the empirical results, the highest model's accuracy is the construction of the neural network, 85.39% of accuracy is obtained.
