A Statistical Study On Improving The Efficiency Of Non-performing Assets Collection In The Background Of Internet Credit Rating System

Internet credit is a new credit business,the use of the user in the Internet activities generated by a large number of data to a more comprehensive assessment of the user's credit level and timely reflect the user's credit level changes,which is the traditional credit evaluation system can not But the Internet user behavior data has a complex behavior dimension and a huge amount of data,and we can only select some of the behavioral dimensions and their data when we make a credit evaluation to the user.Therefore,in the background of such Internet credit,to explore and study how to use the Internet user credit information to establish a personal credit system and how such an evaluation system can reflect the user's true credit level and other issues,the efficiency of non-performing assets Of the increase has a significant and urgent practical significance.The definition of non-performing assets,in the traditional sense,refers to the use of all means and methods within the scope permitted by law,the value of the assets realized and the value of the promotion activities.Since there is a personal or institutional debt,there is debt collection and disposal of this industry,which is a traditional industry with thousands of years of history.Many people generally believe that the disposal of non-performing assets is through constant pressure to force the debtor to make repayment action,which is the traditional bad disposal left a deep impression on everyone,but in the "Internet +" under the disposal of non-performing assets need to use the new platform deal with.Small bad debt is characterized by a huge amount but a single amount of small,and most unsecured,covering a wide range,with the help of the Internet and the power of large data in order to achieve efficient disposal,can not rely solely on human sea tactics.On the other hand the new financial debtors also have new features,they are complex background,but are all networked survival of the crowd,whether it is 80 after 90,their life,consumption and lending behavior mostly occurred online,through a large number of lines On the data analysis,such a crowd to conduct a more clear description of the accurate positioning,which the late collection of work can play a great support and help.The need for data-based analysis and intelligent matching system to match;and through the analysis of large data on the creditors for more in-depth mining and portraits,the understanding of the debtor makes the means of disposal of non-performing assets richer,more humane approach,The process of disposal is more legal,sunny,more efficient disposal.This paper is based on the following logic: first analyze the factors that affect the efficiency of non-performing assets collection,which depends on the three factors: First,the difficulty of the case itself,that is,the debtor bad credit after the collection of the difficulty of collection;The company's reminder of the possibility,depending on the collection methods,the enthusiasm of the collection staff and communication skills;Third,the matching of the case,that is,the company platform for the upstream side of the case and the downstream side of the company's distribution is reasonable and optimal,Depending on the rationality of the distribution of cases in the pool.Second,there are two factors that affect the debtor's bad credit: one for the debtor's own factors such as gender,age,education,location,occupation,income level,whether the car,marriage and other indicators;the other for the Internet consumption data Such as the total amount of arrears,overdue arrears,the purchase of goods,the number of arrears,the previous loan records.According to the author's internship company in the data obtained from the database,the greatest degree of mining to the debtor more comprehensive information,including text mining on the Internet consumer goods,through the relevant statistical methods and ideas on the debtor in all directions Image,the debtor's return rate to predict,more easy to pay back the crowd to give priority to the collection work,so as to effectively improve the overall efficiency of the collection.In the process of the debtor's portrait,the author attempts to use K-means clustering,regression analysis and machine learning methods,and through the training set and test set classification of the model to test and optimize the KS test and ROC curve comparison The optimal model to predict the debtor's return rate.In the actual work,the author constructs the model through the decision tree classification method has been applied to the actual collection work and the collection efficiency has been significantly improved.I hope that through the Internet technology and the joint efforts of researchers,from the debtor's point of view to help him deal with non-performing assets,as much as possible to reduce the debtor's debt pressure,and even through the docking of various physical assets or human assets realized platform,help The debtor to enhance the ability to repay,build a better "Internet +" bad asset disposal platform to provide more information on Internet credit.