As a multiple criteria decision making (MCMD) technique, the technique for order preference by similarity to ideal solution(TOPSIS) traditionally has been applied in multiple criteria decision analysis. Based on D.Wu's data mining model, the TOPSIS model presented in this paper has improved from two aspects. Firstly, it extents to deal with both crisp and fuzzy data; Secondly, in order to really following automatic machine learning principles to the largest extent, the weights must be immune to the subjective element and the data noise. Here, the weights are obtained from data sets based on support vector regression(SVR), which is a more robust and efficient data regression method than the traditional data regression method. Thus the proposed model can provide additional efficient tool for comparative analysis of data sets. We apply it in supply chain complexity evaluation, and simulation is used to validate the proposed models.