Personal Credit Evaluation System Based on Decision Tree

Abstract

This paper analyzed personal credit evaluation system in rural area based on decision tree model. Firstly, it reviewed some references about credit evaluation methods and found that decision tree was a linear adaptive data-driven model with induction ability and a wide range of function approximation ability so that it could be applied into personal credit evaluation. Secondly, decision tree classified data samples consisting of two phases: constructing decision tree model and then classification stage. The first stage was to train data samples to establish a decision tree, and this process was divided into three steps which included feature selection, node splitting and tree pruning. The second stage was to put test samples into the established decision tree, and let it to classify from a new set of data. After that, it took advantage of the model to evaluate personal credit and selected the twenty indicators. The results showed that household assets, net assets and the existing current account balance were the most important three indicators for evaluating personal credit in rural area.