Research on Combined Model for Evaluation and Prediction of Zero-Sum Game Systems Based on EWM-TOPSIS and RBF-Multiple Regression

This paper proposes a combined evaluation prediction model based on EWM-TOPSIS and RBF-multiple regression for the analysis of zero-sum game systems. The model first calculates the objective weights of each evaluation index by entropy weight method (EWM) to reduce the influence of subjective factors; and then uses the TOPSIS method to rank the evaluation objects to obtain more reliable evaluation results. In the prediction stage, radial basis function neural network (RBF) is introduced for nonlinear modelling and combined with multiple regression analysis for trend prediction to improve the accuracy and stability of prediction. Our model results performs well in terms of evaluation consistency, prediction accuracy, and computational efficiency. Compared with deep learning methods, it is more advantageous in the case of limited computational resources.