The integrated method to determine weights based on the divergence of attributes

The standard deviation or the entropy measure is usually used for quantifying divergence degree and deriving objective weights of attributes. Considering the conventional entropy weight method may create great gap among different weights, this paper proposes the improved entropy weight method to decrease the unreasonable gap. Moreover, in order to improve the sensitivity of weights, the standard deviation method based on the reciprocal judgment matrix is proposed, and the weights derived from the two methods are integrated according to the least square principle. In the end, a numerical example is used to illustrate the stability and applicability of the proposed method.