Research On The Methods Of Financial User Profile And Credit Evaluation In Cloud Computing Environment

In recent years,with the rapid development of the Internet,people's lives have undergone significant changes.Riding on the east wind of the Internet,Internet finance also has room for vigorous development.The limitations of the previous traditional finance have been significantly improved,and financial informatization has become an irreversible trend.The financial risk challenges that accompany financial informatization have made us realize that improving the security of financial development is an important issue.User portrait is to divide user groups with different characteristics by analyzing the user's financial data.Credit assessment is to use big data to establish a risk control model,and use the model to determine whether the borrower may default,thereby protecting the interests of third-party lending platforms and ensuring the orderly and healthy development of the industry.This thesis studies the financial user profile and credit evaluation methods in the cloud environment.The main innovation contributions of this thesis are as follows:(1)A method for classification of financial user portraits based on feature weighting is proposed.The TF-IDF improved algorithm(TF-IDF-CF)is used to calculate the different weights of feature attributes in the classification process,and then the naive Bayes based on feature weighting is used.The Sri Lankan classification algorithm divides the user group,analyzes the profile of different groups according to the divided groups,determines different types of users based on the degree of contribution of financial users to financial institutions,and obtains user level information.Experimental results show that the algorithm can efficiently and accurately achieve user level division when the amount of data is small in a stand-alone environment.(2)A parallel method based on feature weighting of financial user profile classification is proposed.The improved algorithm is parallelized on the Map Reduce framework to realize the processing of millions of financial user data in a short time and eliminate TF-The inter-class deviation of the IDF algorithm is weighted into the naive Bayes algorithm model.Makes some of the more important attributes in the financial industry get reasonable weights.When the amount of data is large in the cloud environment,the algorithm has accurate classification and good stability.(3)A method for classification of financial user credit evaluation based on feature selection is proposed,preprocess the data,use the CSAFS feature selection algorithm to perform feature selection,which can effectively reflect the behavior information of customers;then use logistic regression,Xgboost,Catboost three models are used for predictive analysis,and the classification performance of the model is evaluated.Experimental results show that the algorithm can efficiently and accurately evaluate the quality of user credit when the amount of data is small in a stand-alone environment.(4)A parallel method of financial user credit evaluation classification based on feature selection is proposed.The improved algorithm is parallelized on the Mapreduce framework to realize the processing of millions of financial user data in a short time.After using the existing financial user data and verifying the logistic regression model,Xgboost model,and Catboost model,the results show that the parallel method is effective and feasible in user credit evaluation and has good stability.