Most of the existing credit evaluation index weight estimation models only consider the cross-sectional data of a single time point, and do not consider the characteristics of the index data change over time, and the obtained index weights cannot reflect the objective data change ability of multiple time points in the sample, so this paper proposes a new weight calculation model based on the dynamic assembly weight of particle swarm optimization (PSO) algorithm and empirical analysis on Family Farms and Ranches (FF&R) in Inner Mongolia. Firstly, the weight of a single point static credit evaluation index with default identification ability is measured by the Fisher discriminant method, and secondly, a nonlinear programming equation is constructed to dynamically assemble the weights of each single time point by minimizing the overall deviation of the dynamic assembly weight of each time point weight, and the assembly weight of credit evaluation indicators that can reflect the data change ability of each time point is measured. Finally, the comprehensive credit evaluation score of each sample is calculated by linear weighting. The innovation point of this paper is to construct a nonlinear programming equation with the smallest sum of squares of the deviation between the weights of each time point and the assembly weight, and find the dynamic assembly weights that accurately reflect the change ability of data at each time point, which makes up for the disadvantage of ignoring the continuity of time in the traditional credit evaluation weight measurement method and not obtaining the weight of evaluation indicators reflecting the change ability of multiple time points.
本站僅提供期刊文獻檢索。
