Application of Neural Network Based on Flexible Neural Tree in Personal Credit Evaluation

With the development of society and economy, the business scope of banks is constantly expanding, especially the credit business is getting stronger and bigger, and then the credit evaluation problem appears. We use the Flexible Neural Tree-Neural Network (FNT-NN) method to classify bank customer credit card default data and find that a neural network constructed based on a flexible neural tree can achieve better classification results. FNT-NN is based on the FNT solution to construct a neural network and use Back Propagation (BP) neural network algorithm to optimize the connection weights. The flexible neural tree uses a tree structure coding. The parameters are usually optimized using the Particle Swarm Optimization (PSO) algorithm. It can find a globally optimal network suitable for the data set. By comparing logistic regression, support vector machine and decision tree, the experimental results show that FNT-NN can get better results.