Credit scoring card characteristic selection method based on Monte Carlo search

The invention discloses a credit scoring card characteristic selection method based on Monte Carlo search. The method comprises the following steps that: in a credit risk rating scene, obtaining credit scoring characteristic data; initializing a Monte Carlo search tree; selecting a node which is not completely expanded, and using a partial regression square and an index to carry out compound strategy extension; on the basis of an expanded tree structure, importing the partial regression square into a random simulation strategy, and dividing a random selection domain to generate a simulation path; on the basis of characteristics selected by each path, independently using a cross verification method to carry out logic regression modeling, and calculating a relevant statistical magnitude; and on the basis of a UCB (Upper Confidence Bound) formula, selecting an optimal path. By use of the credit scoring card characteristic selection method which is put forward by the invention, the modeling effect of a credit sco