K-means and RS based DEA model and its application in Chinese low-carbon efficiency

Abstract

Abstract Since the evaluation results of DEA method are difficult to reflect the differences among many samples, the rough set idea is introduced, the comprehensive efficiency value solved by DEA is taken as the decision attribute, and other indicators are taken as the conditional attribute, and the index is relatively reduced. On the one hand, the kernel attribute is extracted to reduce the redundancy among indicators and maintain the unity of the index system. On the other hand, the provinces with effective and invalid DEA results are segmented into low input and high output, high input and high output, low input and low output, and then the low carbon efficiency of different provinces can be analyzed and the solution direction can be proposed. However, the data used in rough sets must be discretized data. In this paper, the contour coefficient method is used to optimize the K-means algorithm to obtain the optimal cluster number and classification results, so that the discretized data obtained can be used for the relative reduction of rough sets. The results show that Anhui, Hunan, Jiangxi, Chongqing, Guizhou, Shaanxi and Heilongjiang with invalid DEA show low input and low output and high input and low output. The provinces with effective DEA and high output and low input and high output are Shanghai, Gansu, Qinghai, Yunnan, Xinjiang, Henan and Zhejiang with the best low carbon efficiency. The provinces with high input and low output of DEA are Beijing, Hubei and Sichuan, which do not allocate resources well and have the worst low-carbon efficiency. On the whole, the distribution of low carbon efficiency has spatial autocorrelation, and there are high-high aggregation, high-low aggregation and low-low aggregation.