企業能夠適當的分配投入與產出將有助於企業本身穩定的成長，但企業本身對於自身績效的評估往往無法確切的評估，因此本研究期望使用一個合理的評估機制來衡量台灣上市櫃電子公司績效情形，為產業間帶來比較的結果，使其得以認識產業間績效優良企業，並學習改善自身的資源分配。本研究使用資料包絡分析法來衡量企業之公司績效，運用此方法進行績效的評估，結果將會評比出績效優良的公司，為經營無效率的企業提供經營策略改善學習之方針；而同時運用核心主成分分析法做為簡化變量的方法，在眾多的投入與產出變數中選擇具指標的變數，資訊的維度縮減將會改善資料包絡分析法投入與產出項過多時所造成的共線性問題，而本研究所採用核心主成分分析法在對於非線性資料的處理有著極佳的處理效果，強化非線性資訊解釋能力，並且經由分析過後，使得績效評估結果決策單位間的平均效率值大幅降低，造成評比的級距更加明顯，得以確立決策單位間效率之差距，區隔出效率優良與低落的企業，藉此更加精確改善公司績效的評估結果。
Appropriate allocated corporates’ resource on suitable objects can increase the possibility of sustainability. However, how to precise assess the corporates’ operating performance is a critical problem. Thus, this study proposes an hybrid architecture for corporate to measure the operating performance suitably. The architecture can help managers to pick up the corporate with superior operating performance. The corporate with inferior operating performance can follow or learn form the corporate with supe-rior operating performance. The performance measure is implemented by data envel-opment analysis (DEA). The method can handle multiple inputs and multiple outputs as the same time. However, too many input variables and output variable will deteri-orate the discriminant ability of DEA. Thus, this study performs kernel principal component analysis (KPCA) to handle the problem of dimensionality so as to in-crease the discriminant ability of DEA. In addition, most proportion of real-world data were non-linear structure. The traditional linear technique such as principal component analysis (PCA) can not handle it well. According to our research finding, the original inputs variables and output variables undergoes the KPCA and then fed into DEA will increase the discriminant ability. Thus, the hybrid archtechture will help investors to identify the corporate with superior operating performance so as to modify the operating strategies to survive in highly turbulent environment.
