| With the continuous development of China's economy and the continuous improvement of the stock market system,stock investment has become a common investment behavior,but investors are affected by their own knowledge and information oligarch,which makes the return on investment lower.Therefore,how to accurately identify stock price fluctuations and make correct investment decisions to improve stock investment returns has become a matter of great concern to the industry,academia,and many investors.This article also attempts to choose some effective indicators and models from this perspective.Construct a model with high prediction accuracy to assist investors in making investment decisions.In the past,there have been a large number of studies that predict the stock price change on a single trading day.This article improves this and chooses to predict the stock price change over a period of time,that is,the increase in the stock price over a period of time relative to the base period(the highest increase)Is the predicted variable.The number of ups and downs in China's stock market indicates that China's investment market has a higher speculative risk,which means that changes in stock investor sentiment and their attention to stocks may cause changes in stock prices.Among them,investor sentiment and the degree of attention to stocks determine his investment decisions to a certain extent,and then cause stock market volatility.With regard to the selection of indicators,because the two over-the-counter factors,investor attention and investor sentiment,will affect investor decisions to a certain extent and affect stock price volatility,this article intends to construct investor sentiment and attention indicators based on Internet data.In this paper,two out-of-field factors,investor attention and investor sentiment,are added to a machine learning algorithm to see if it can improve prediction accuracy.Based on this,a high-precision prediction model is built to assist investors in making investment decisions.First of all,this article uses Oriental Fortune to obtain the hot debate words of the Shanghai and Shenzhen 300 Index,and then uses some of the hot debate words as keywords to download its corresponding Baidu index,thereby constructing an investor attention index.Secondly,some keywords were used asWeibo search keywords to capture Weibo text content,and sentiment analysis was performed on the obtained text to construct investor sentiment indicators.In terms of model selection,with the continuous improvement of computer processing performance,artificial intelligence prediction models based on intelligent algorithms such as neural networks and support vector machines have been widely used in financial asset price prediction.And the advantages of complex information processing,with the help of computer software,make the prediction of stock price fluctuations more accurate.Because this article explores a binary classification problem,three commonly used binary classification prediction models(Decision Tree Model,Support Vector Machine Model,and BP Neural Network Model)are selected for research.Based on the above,this article adds these two indicators to the three binary classification prediction models.The research results show that the models with both indicators have higher accuracy and precision,and the model's prediction performance is even better when investors' attention indicators are added.At the same time,in order to obtain higher prediction accuracy,this paper applies the AdaBoost lifting algorithm to these three prediction models,that is,the AdaBoost lifting algorithm uses the 3 prediction models as weak classifier models to weight fusion and form3 learning capabilities.Strong predictive model.The results indicate that the AdaBoost lifting algorithm has optimized these three prediction models to a certain extent.The performance of the decision tree-AdaBoost lifting model is optimal,and its prediction results have a greater effect on assisting investors in making investment decisions. |
PDF Full Text Request