Research on the selection of evaluation index in service management

In comparison with the single attribute evaluation, multi-attribute evaluation is to given an overall assessment of the effect and performance of the system by using several indexes and comprehensive evaluation function. With the number and forms of the evaluation indexes increasing, how to select the evaluation indexes is very important before evaluation. This paper mainly solved the selection and eliminating correlation of evaluation indexes, also known as index reduction. The methods of calculating the correlation degree of linear synthetic index and nonlinear synthetic index are given. Then a dynamic index reduction algorithm based on equivalence class partition is proposed. In many cases, the index reduction is occurred in an equivalence class, so the algorithm can reduce the range of index reduction. It also can make full use of the anterior reduction results to increase the efficiency of algorithm. The simulations experiments show that the more indexes, the more efficient this algorithm can be.