A clustering method for evaluating the environmental performance based on slacks-based measure

• SBM measure was used to classify the environmental performance of Chinese industry.
• The context-dependent DEA method was used to get the sub-clusters for detailed managerial meaning.
• Our approach can avoid the non-disjoint property in Po et al.’s CCR-clustering.
• Compared to k-means clustering, our approach is more proper to deal with input-output production feature.

The conventional clustering algorithms are mostly distance-based, which can lead to distorted results in the evaluation of production unit’s performance. As a non-parametric method, data envelopment analysis (DEA) has become a popular approach to measuring the production process performance. However, few researchers paid attention to the relationship between clustering approach and DEA. In this paper, we use a non-radial DEA framework (slacks-based measure, SBM) to classify the environmental performance of Chinese industry, forming a benchmark-based clustering approach. Additionally, we employ the context-dependent DEA method to get the sub-clusters for detailed managerial meaning. An application in real world is given to explain the usage and effectiveness of the proposed SBM-based clustering method, and the result is compared with the conventional distance-defined k-means clustering approach.