Stock Recommendation Method Based On Mixed Factors

China's stock market has been established for more than 20 years.During these two decades,the stock market has continued to grow and develop,and stock market participants have experienced explosive growth.According to statistics,China's stock market is dominated by retail investors,and most of the retail investors' expertise is weak.By reviewing previous studies on stock recommendation methods,it is found that there are two main types of current stock recommendation methods: the recommendation method based on stock evaluation and the recommendation method of stock price forecast.With the development of information technology,technical analysis methods such as machine learning have also been applied to the field of stock recommendation,and some achievements have been made.In order to improve the status quo of China's stock market,correctly guide the value orientation of stock market investors,and reduce the losses caused by blind investment,this paper proposes a practical recommendation of retail investors as a starting point,and proposes a stock recommendation method based on random forest model.Recommended for stocks with higher yields than the Shanghai Composite Index.Summarizing the research results and shortcomings of the previous recommendation methods,this paper is different from other stock recommendation literatures in the selection of dependent variables and independent variables,and the selected indicators are more comprehensive.Selecting the listed companies in the real estate industry as the research object,based on the combination of securities analyst rating,analyst forecasting characteristic indicators,listed company's basic information characteristics indicators,the number of securities company analysts and the number of published research reports,and the financial indicators of listed companies,choose 30 features are explored as input variables of the model.First,the data is preprocessed by missing value processing,outlier detection,data conversion,etc.,to prepare the data for the model.The random forest model was selected for modeling and compared with Logistic regression model,AdaBoost model and XGBoost model.And the precision is mainly used to evaluate the pros and cons of the model.The results confirmed that the random forest model has the highest precision,and the XGBoost model has the highest correct rate,and the effect of random forest is far superior to the logistic regression model,which is slightly better than the AdaBoost model and the XGBoost model.