P2P Lending Platform Default Risk Identification Based On Boosting-SVM Algorithm

With the continuous and steady development of China's economy and the continuous penetration of the Internet in people's lives,Internet financial products is the inevitable product of the great era background.P2P network lending is the most representative model of Internet finance.P2P network lending platform as an information intermediary platform,can effectively link the two sides together for the two sides to create value for the borrowing.P2P network lending industry after several years of barbaric growth,the product's comprehensive rate of return gradually return to rational state,the industry's turnover has gradually stabilized,the entire P2P network lending industry has entered a stable development stage.Although P2P network lending platform can create value for both parties to achieve profitability,in fact P2P network lending platform to take a lot of risk,which due to borrowers breach of contract caused by bad debts is the biggest risk to the legal network borrowing platform.The After the stable development stage of the P2P network borrowing industry,the competition between the P2P network lending platform is not only reflected in the products and services,but also the core competition is reflected in the platform's ability to identify the default risk of the borrower.This paper proposes to determine whether the borrower's existence of a breach of contract may lead to the risk of bad debts can be studied as a pattern recognition problem.Through the use of machine learning,artificial intelligence method to fully identify the risk of default debt,can effectively help P2P network lending platform to reduce the risk of bad debts,so as to promote the healthy development of the network lending market to better serve the socialist market economy.Support vector machines are one of the most common classification algorithms in machine learning and can also be used to deal with regression problems.In support of small sample,nonlinear and high dimensional pattern recognition problems,support vector machine learning algorithm shows many unique advantages.Compared with other machine learning algorithms,support vector machine learning algorithm has a solid theoretical basis and simple and clear mathematical model.Boosting tree algorithm is an important part of Ensemble learning.The idea of its classification algorithm is to increase the sample weight for training samples which are easily misclassified and to reduce the number of classified samples by learning as much as possible.The Boosting tree algorithm can construct a series of simple primary prediction methods,and combine them with a certain rule to get a complex and accurate classification prediction method.Aiming at the shortcomings of SVM in dealing with large sample training set and the shortcoming of Boosting tree algorithm in dealing with high dimensional pattern recognition problem,this paper proposes a Boosting-SVM algorithm based on support vector machine and Boosting tree algorithm.Boosting-SVM algorithm uses the reduced support vector machine as the basic classifier in the framework of the Boosting tree algorithm.The sample weights of each training sample are added to the objective function of support vector machine optimization problem.By using the sequence least optimization algorithm The basic classifier is obtained and the basic classifier obtained by each iteration is linearly combined to obtain the final classifier.Adding the sample weight of the training sample to the objective function of the support vector machine optimization problem can make the different sample punishment different,so the different sample weight Samples can get different degrees of attention.When the minimum optimization algorithm is used to solve the problem,the weight of the training sample is added to the objective function of the optimization problem,so the lower bound of the inequality constraint range of the variable is reduced in the solution.In this paper,we use the machine to study the open data set Breast-Cancer data set in the original support vector machine and Boosting-SVM algorithm to verify the Boosting-SVM algorithm to improve the effectiveness.The final use of the experimental data is the use of web crawlers in the P2P network lending platform to collect information on borrowing users.Based on the experimental results,the Boosting-SVM algorithm is superior to the original method in terms of accuracy and computational efficiency.The results show that the Boosting-SVM algorithm is better than the original one in the original support vector machine and Boosting-SVM algorithm.Support vector machine,therefore,Boosting-SVM algorithm can more effectively help P2P network borrowing platform to identify default risk.