Stepwise Regression Algorithm Based On Decision Tree And Its Application In Stock Forecasting

With the continuous development of China's economy,China's stock market is more and more perfect,and gradually as a barometer of China's economy and observe the important indicators of China's economy,many investors have invested on the stock market,hoping to get in the stock market Income,a large numb er of experts and scholars have also engaged in the stock market research and analysis,and summed up a number of stock forecasting algorithms and models to help investors reduce the risk of investment stocks.Stock market every moment will produce a lot of data,these data include the stock market daily stock data,listed companies,financial data and so on.These massive amounts of data seem to have no rules,no connection,but behind the hidden hidden information is not known.Due to the large amount of data,it is difficult to predict and analyze stocks using traditional methods.With the rise of large data,data mining technology has become a hot research data,data mining technology can be from the huge seemingly no law,chaotic data to extract valuable information.Therefore,based on the fundamentals analysis and technical analysis method based on stock analysis method,this paper proposes a stepwise regression algorithm based on decision tree based on stepwise regression algorithm and CART decision tree algorithm in data mining technology.Used in the stock forecast,the A-share listed companies in the annual report of the financial indicators as the object of analysis,the stock forecast and analysis.This paper chooses the financial indicators of 2013,2014 and 2015 as the data sources of 2007 A-share listed companies,and analyzes the 29 representative financial indexes of the seven indexes as the input variables,and uses SPSS Modeler software to establish Related models,70% of the data were randomly selected as the training sample set,30% as the test sample set before the model was established.In the step-by-step regression algorithm and the financial index model of listed companies,the financial indicators with significant impact on earnings p er share are selected,and the accuracy of these indicators on the earnings per share forecast is analyzed.Then,the CART decision tree algorithm is established and the classification of financial indicators of listed companies Forecasting model,find out the relationship between high quality stock and financial index,analyze the prediction accuracy of CART decision tree model,then improve the CART decision tree model,improve the accuracy of model quality and model prediction,and finally establish stepwise regression algorithm based on decision tree The new model first uses the improved CART decision tree model to screen out the important financial indicators,and then sets the selected financial indicators as the stepwise regression input variables to establish the model,compared with the single stepwise regression model,the reduction of the financial indic ators of the listed companies,the interference of unrelated financial indicators on stock forecasting,the elimination of unrelated indicators to mislead investors when analyzing stocks,improved forecast accuracy,and helped investors reduce the risk of stock investment.