基于创新算法的理财产品个性化推荐

北京理工大学珠海学院2020届本科生毕业论文Personalized Recommendation of Financial Products Based onInnovative AlgorithmsAbstractIn the era of big data,the problem of information overload is becoming moreand more obvious,so it is particularly important to innovate and develop recom-mendation algorithms to mine more valuable information.The data in this papermainly depends on the service data of santander's customers from 2015-01-28 to2016-05-28,with a total of 13647309 service records.Through data cleaning,fillingin missing values,deleting duplicate records and descriptive statistics,the data werepreliminarily explored.User characteristics were extracted by filtering method,cus-tomer portraits were established on the basis of feature engineering and purchasebehavior labels were given to customers.By Logistic regression study to find outthe characteristics of customers who buy annuities.Secondly,by K-Means cluster-ing for the data points to divide customers into active,inactive and three types ofpotential state,thus trend model based on BP neural network to establish customerservice come to the conclusion that accuracy of 96.44 %Finally,through the naivebayes algorithm in June 2016 to the customer going to use the products of person-alized recommendation,its accuracy is 96.52 %and based on the conclusions,somereasonable suggestions are presented.Keywords:Recommendation algorithm BP neural network K-Means clusteringCharacteristics of the engineering Logistic regression Nave Bayes