A New FA-DEA Method Based on Classification and Weighting

In order to avoid the drawbacks of current FA-DEA method, this paper put forward an improved method based on classification of original indexes to extract new indexes according to the weighting of variance contribution, which has considered the information redundancy when there is high correlation of indexes, abnormal value of similar indexes, as well as the difference of each common factor in information content. Moreover, it can also help to avoid information loss, which is comparatively more reasonable. We use large sample and long-term data from holistic listing samples in Chinese security market to examine different kinds of methods. Our results show that the new method based on classification and weighting offers advantages over the traditional ones.