Mining And Analysis Of Stock Market Public Opinion Data

With the rapid development of computer technology and the explosion of network data information.Data mining technology has gradually become the core force to promote the innovation and development of financial stock market in the information age.In the stock market,more and more investors like to share their experiences in the online stock forum.For the above reasons,a large number of public opinion data of stock evaluation on the network has become an important factor affecting the development of the stock market.In various studies on the relationship between networkt stock comment data and stock market,the traditional single structured stock market trading index has been unable to meet the needs of people to analyze and predict the situation of the stock market.This paper on the basis of machine learning and deep learning technology deeply studies and analyses the stock comment data of network stock market,and tries to solve the problem of insufficient ability of traditional stock market forecasting methods and emotional classification of stock comment data.The relevant research contents are as follows:1.The crawler technology is used to obtain the commentary information and other specific attribute list information of Eastmoney Web according to specific rules,and it is used as the stock comment data.Using various tools to Pretreatment the original data,such as de-noising,de-interference,word segmentation,stop word filtering,etc.In the process of structuring Chinese word segmentation and feature matrix construction,TF-IDF technology and Word2 vec are introduced to complete text structuring,which improves the learning effect of classification and prediction models.2.Taking the emotional tendency of mining stock market public opinion data as the research objective.Two classifiers based on Naive Bayesian Classification and Convolutional Neural Network are established to classify sentiment tendency of stock comment data.The classification results of the two classifiers are compared and analyzed according to the classifier evaluation index.The results show that,the accuracy of sentiment Tendency Classification Based on CNN is higher than that based on Naive Bayesian sentiment classifier.The study further proves the accuracy of CNN in classifying public sentiment tendency of stock evaluation in the field of stock market.3.This thesis studies the influence of structured weight intensity of hidden emotional information on stock market volume in stock comment data.The results show that the emotional value of stock reviews calculated by matching with the emotional dictionary can predict the volume and its changing trend to some extent.4.SVR model and BP neural network are used to study the impact of stock review data on yield.The performance evaluation indexes of SVR model and BP neural network are compared and analyzed.The results show that both models can predict the stock market return and its trend,and the prediction error of SVR model is less than that of BP neural network,which further verifies that SVR model can produce better prediction results than BP neural network.