Desing And Implementation Of The Pension Portfolio Investment Recommendation System Based On Machine Learning

In recent years,with the development of science and technology and the improvement of medical standards,the number of aging population in my country has increased.This phenomenon will be a problem faced by our country for some time to come.Pensions are an important source of funds to protect the basic living standards of retired people.There are three sources of pensions.The first part is the pension paid through the fund pension insurance system,the second part is the supplementary pension based on the enterprise annuity or occupational annuity,and the third part is the personal savings commercial pension insurance.The first and second parts are compulsory pension payments,while The third part is at the discretion of the individual and can be used to invest in different types of financial products in order to obtain benefits.The problem of old-age care is no longer only for the elderly to consider the problem,in the trend of continuous upgrading of consumption,everyone's requirements for quality of life are also constantly improving,to a certain extent,pension can not guarantee the quality of life after retirement.Early investment for the aged is something everyone needs to do.In this article,people who are willing to invest in retirement and are willing to invest in retirement are collectively referred to as individual retirement investors,or investors for short.At present,there are many investment and wealth management products on the market.For those investors who are interested in investing but have no professional knowledge and energy,how to choose their own wealth management products is the biggest problem encountered by many people.Many wealth management platforms recommend the same wealth management products to investors,and none of them is a platform specifically for pension investment.Choosing a product that is not suitable for you will bear a relatively large risk,which is contrary to the original intention of investment.Therefore,this article uses machine learning to create a recommendation model and construct a recommendation system for pension products.Using scientific methods,recommend to each investor the pension and wealth management products suitable for him.The main research contents of this article are as follows:(1)The behavior data of purchasing wealth management products by paying attention to the history of investors,including the transaction flow of investors purchasing wealth management products and the flow of investor funds,creating models for investors to solve the unfriendly impact of blind recommendation on investors and wealth management platforms,To improve the success rate of pension portfolio financial product recommendations.(2)The intersection of chi-square test and random forest screening results is used as the final feature screening result in feature screening,which can effectively prevent the limitation caused by using only one method.(3)Adopt XGBoost(e Xtreme Gradient Boosting)algorithm,use investor basic information and historical purchase wealth management product data,including basic property information of wealth management products,investor transaction flow,investor capital flow,number and amount of wealth management products held by investors,etc.,to predict the risk level of investors.The level of risk predicted by the actual behavior of investors is more accurate than the traditional questionnaires or online risk assessment methods.The recommended pension portfolio wealth management products are more suitable for investors.(4)The user-based collaborative filtering algorithm is used to recommend pension portfolio products to investors who can not obtain the evaluation results recently or whose evaluation results have expired and have not been re-evaluated,so as to make up for the lack of accurate recommendation using the algorithm model when the basic information of investors can not be obtained.The pension system recommendation system studied in this paper aims to solve the inability of investors to choose wealth management products.Similarly,for wealth management platforms,for investors who come to consult,they have their accurate investment preference analysis,which effectively guides investors' investment behavior and improves the recommended Success rate.