A study of algorithms for solving nonlinear two-level programming problems oriented to decision tree models

Abstract

In this paper, the original two-level planning problem is transformed into a single-level optimization problem by combining the penalty function method for the large amount of data processing involved in the training process of the decision tree model, setting the output as a classification tree in the iterative process of the CART decision tree, and recursively building the CART classification tree with the training set to find the optimal solution set for the nonlinear two-level planning problem. It is verified that the proposed solution method is also stable at a convergence index of 1.0 with a maximum accuracy of 95.37%, which can provide an efficient solution method for nonlinear two-level programming problems oriented to decision tree models.