The Research On Personal Credit Evaluation Mode Base On BP Optimization Algorithm

The rapid development of the market economy has prompted consumer credit which is different from traditional consumption patterns becoming a new trend. Personal credit is a basis of consumer credit evaluation, which reflects the ability of the person to fulfill credit commitments. Now it is no longer a virtue, more precisely, is a commodity, is an ability to pay on the basis of personal net assets and expected future revenue. The use of advanced technology to build personal credit evaluation model, and then objectively correct evaluation of personal credit can be fair to safeguard the legitimate interests of both the credit side and the fiduciary body, but also provides an effective tool for credit institution to predict credit risk of credit institutions, and at the same time it maintain a healthy and stable development of market economy.Based on the analysis of the status of the domestic credit market and the research at home and abroad in the field, the paper introduces two kinds of artificial intelligence algorithm, BP neural network algorithm and particle swarm optimization algorithm, and then optimize the personal credit evaluation index system to Germany personal credit data set by use of the attribute information entropy, compare the gain value index to select the suitable credit evaluation indexes, so as to construct a new index system. Secondly, create two suitable comprehensive evaluation index to measure comprehensive performance of the personal credit evaluation model: comprehensive evaluation of performance index and loss rate of comprehensive evaluation index. Besides, according to the index system combined with the new BP algorithm and PSO algorithm to construct the BP routine optimization of different training functions of personal credit evaluation model and the evaluation of personal credit model based on PSO optimization. The simulation result shows that: in the credit evaluation models of conventional BP optimization algorithm, the trainbfg training function model is with faster convergence speed, and has a good generalization performance, the two comprehensive evaluation indexes also indicate that the trainbfg training function credit model has the best generalization performance among similar credit evaluation models based on conventional BP optimization algorithms. Under the same conditions, the comprehensive performance of PSO optimization model can change with the number of iterations, the experimental conclusion show that the comprehensive performance of PSO optimization model is better than the model based on conventional BP optimization algorithm when the iteration number is 300.In recent years, with the rapid development of computer technology and artificial intelligence technology, more and more experts and scholars used the technology to build a credit evaluation model, and they had also made a lot of research results in the field. This article aims to evaluate the performance of the various models based on the summarized results of previous studies. Then use examples to validate the superiority and feasibility based on individual credit evaluation model of artificial intelligence algorithms for subsequent lay-depth study basis, but also have some reference value and theoretical significance for the late improved credit evaluation model.