Research On Prediction For Return Rate Of Bank’s Financial Products Based On Financial News Text Mining Technology

As a part of the asset management market,the bank financial product market plays an important role in the financial field and the return rate will be different in different economic development periods.In April 2018,the "New Regulations on Asset Management" put forward the essence of bank financial products--"entrusted by others and managing money on behalf of others".With the popularization and development of AI,financial news can bring the latest information to investors at the fastest speed.Rational investors will analyze the important information scientifically and make the most favorable decisions,on the basis of taking certain risks,realize the increase of wealth.Text mining technology can be used to process financial news data and predict bank financial returns accurately and quickly,which can provide important information support for investors to make better decisions.Therefore,no matter in the academic field or the bank financial product investment field,the research of this paper has the vital practical significance.In this paper,big data and machine learning methods are combined.Firstly,according to the national policy guidance,economic uptrend and economic downtrend are selected,and the text data of financial news during 2020-2022 are obtained by web crawler.Secondly,Python software is used to do data cleaning,Chinese word segmentation,keyword and word cloud analysis on the obtained financial news text,and then process the Baidu index corresponding to the keywords,data capture and data pre-processing on the yields of several financial products from Bank of Guiyang.Finally,SVM algorithm,random forest algorithm and XGBoost algorithm are used to build the prediction model of bank financial product yield,this paper studies the forecasting effect of financial news text on the return rate of bank financial products under the three algorithms of economic up-phase and economic down-phase respectively.Through the research of this paper,the following conclusions are drawn:(1)There is a certain correlation between financial news text and the return rate of bank financial products.The return rate prediction model of bank financial products based on financial news text has achieved good prediction effect in both economic upturn and economic downturn.(2)The prediction effects of the three algorithms selected in this paper,SVM calculation,random forest and XGBoost,are different in different periods of economic development.XGBoost algorithm is the optimal prediction in the uptrend of economic development,and XGBoost algorithm and random forest algorithm are the optimal prediction in the downtrend of economic development.(3)In different periods of economic development,the XGBoost algorithm prediction model constructed in this paper can effectively predict the return rate of bank finance,and the algorithm can be effectively applied to actual investment activities.