A hybrid modified DEA efficient evaluation method in electric power enterprises

With the increasingly fierce market competition, efficient evaluation plays a more and more important role in the development of many enterprises. Conducting efficient and inefficient analysis in major enterprises can help each enterprise grasp the changing trends of future development, and better formulate the gap between the production frontier and itself, thus can better know how to make up for the deficiency and be competitive. Data envelopment analysis (DEA) is often utilized to assess the efficiency and attain an envelopment curve. However, the traditional DEA methods have some shortages in dealing with desirable outputs and undesirable outputs. Besides, its time-consuming to solve lots of constraints when new data come. To overcome the insufficiency and complexity, this paper aims to provide an effective solution by modifying DEA methods and integrating data mining algorithm. By this combination, a simple effective method is proposed to evaluate the efficiency of each enterprise, which is subsequently helpful to transform unsupervised learning problem into supervised learning problem at the same time. Finally, the hybrid method has been conducted in electric power enterprises with the evaluation of raw data from different perspective. Two frequently-used classification algorithms have been employed to illustrate the feasibility of the proposed approach.