The Application And Research Of Data Mining In Stock Analysis

Since the policy of reform and opening up has been implemented, China’s economy is developing rapidly, accordingly the stock market came into being. China’s stock market has experienced ups and downs since its inception and become mature gradually, which makes more and more people intend to invest their money in the stock in order to obtain more substantial return. However, due to the lack of expertise and the information asymmetry, people are investing blindly and speculatively, which makes it difficult to obtain substantial benefit. Much stock data appears in the market, and the listed companies publish much information regularly, so how to effectively use the data to reduce the risk of investor’s investment in order to obtain more benefits has become a question which is very worthy of study and analysis.Large number of domestic and international empirical results showed that the regularly published financial reports contain much information which is helpful for investors to better estimate the intrinsic value of the stock. For the mid-and-long term investors, it is very important to know how to make use of the information to estimate the future investment value of the stock. The paper aimed to analyse the internal relations between the financial data that published by listed companies and the stock investment value to mine useful information and make better estimation for the investment value. The traditional statistical models need strict requirements for the data, and they need excessive assumption over the data, which is almost impossible to meet in practical data. However, the data mining technology requires comparatively low condition and it is capable of processing non-normal, non-stationary and high noise data. With the combination of statistics, machine learning and artificial intelligence technology, the data mining can perform nicely while processing massive amounts of data and the high-frequency data. In addition, the data mining is able to work on the dynamic update of models for the continuously received data, so it is very adaptable to the new environment. Therefore, the paper tried to analyse the financial data of listed companies by the method of data mining, aiming to find out the connection between the financial data and the stock investment value in order to provide a reference for investors. The methods of data mining that used in the paper include logistic regression model, decision tree classification and the neural network model. The paper tried to find out which financial indicator has a greater impact on the stock price of listed companies by using these three methods, then we estimated and analysed the results, and compared the results got from empirical analysis to have a better estimation for the stock investment value.