| 摘要: | 邇來企業發展過程中，績效評估對於企業經營管理為重要的一環。過去學術上常使用資料包絡分析法(Data Envelopment Analysis, DEA) 針對產業實例進行績效評估。而資料包絡分析法主要是採用實際投入、產出之數據，利用生產邊界之概念進行效率分析。故其使用之數據需相當精確才能進而評估出具有可靠性之評估結果。而蒐集數據的過程產生之誤差可進一步分為隨機誤差與系統誤差兩大類。過去學術研究中，誤差來源往往排除人為誤差只考慮隨機誤差。而資料包絡分析對於不精確性資料之求解僅提出模糊資料包絡分析法與不精確性資料包絡分析法，此類評估方法所評估出之效率值仍為模糊或僅為區間值，對於誤差數據所造成之評估影響程度並無深入研究。因此，本研究以資料包絡分析法為基礎，於CCR模式與BCC模式下，使用國際貨櫃港埠實例設計一系列誤差方案進行測試，並利用生產邊界示意圖呈現其誤差影響程度。在求解方法上，利用C++程式語言配合數學規劃軟體CPLEX進行模式求解，縮短使用DEA軟體之求解時間。評估結果顯示，效率值等於1之效率良好受評單位受誤差影響下，其影響結果可分為不受任何誤差影響之相對穩定型效率單位、於誤差影響較大時始受影響之效率單位及誤差存在即受影響之效率單位，此三種效率單位可利用A&P模式進一步驗證。而效率值落於0.4~0.7間之效率單位屬於相對敏感之效率單位，其中包含特殊型效率單位，即其誤差程度遠超過於其他受評單位。此種受評單位可利用本研究之測試方式進行篩選，並進一步針對此類特殊型效率單位及相對敏感之效率單位投入更多人力與成本進行數據比對以確保評估結果之準確性。本研究之研究成果可供後續使用資料包絡分析法之研究學者做為參考。The performance evaluation is an essential issue in the enterprise’s operations. In past studies, the Data Envelopment Analysis (DEA) is usually utilized to evaluate the performance of the industry. The DEA mainly focuses on employing the actual input and output data, coupled with the concept of production bound, to analyze the efficiency. However, to accurately evaluate the performance, the acquisition to precise input data is very critical. In addition, the deviation can be divided into two sorts. One is the stochastic deviation resulting from the affection of actual operations; the other is the systemic deviation arising from the influence on the human and the instrument. In past DEA studies, the systemic deviation is usually neglected and the Fuzzy DEA or Imprecise DEA is generally employed to evaluate the performance, causing that the evaluated performance has the characteristic of fuzz. In past DEA studies, the systemic deviation is usually neglected and the Fuzzy DEA or Imprecise DEA is generally employed to evaluate the performance, causing that the evaluated performance has the characteristic of fuzz. Therefore, in this study, we design seven different deviation scenarios, utilizing the real operation data from the international container port, to evaluate the performance, under the CCR and BCC models, with the DEA. To clearly present the change in deviation under different scenarios, we utilize the diagram of production bound to express these changes. In addition, to shorten the solution time for using DEA Solver to solve the test problems, we employ the C computer language, coupled with the use of the mathematics programming solver, CPLEX. The results show that under deviation influence, the decision making units with the efficiency value 1 can be categorized to three different sorts: relative stability units, units influenced by obvious deviation and units influenced by existence of deviation. These decision making units can verify using A&P Model. Other decision making units with the efficiency value from 0.4 to 0.7 are relatively sensitive. It’s contains a special type of decision making units, that is, its deviation of efficiency is much higher than others. However, these sorts can be explored by the proposed method in this study. According to the results, we should spend more human resources and cost on the relatively sensitive units and the special units in order to make sure the accuracy of data and to avoid the impact on deviation. It is expected that the results obtained from our study could be useful reference for the related carriers or studies. |
