Research On K-means Clustering Markov Chain Method For Forecasting The Trend Of Individual Stock

With the rapid growth of financial markets,stocks have gradually become a very important form of financial management.However,due to changes in policies,markets,etc.,investors are often unpredictable about the changes in the stock market,and may suffer significant losses while potentially gaining profits.Therefore,it is of great significance to provide a proven stock price forecasting method.This topic combines the research of qualitative variables in statistics with the study of mathematical quantitative variables.Using the clustering algorithm in data mining,the Markov forecast model is established to realize the forecast of the price trend of individual stocks.Taking Chinese specific stock as an example,the validity of the method is verified so as to provide the stock investors with more reference plans for stock price prediction.Firstly,on the basis of comparative analysis of many clustering algorithms,this paper selects the k-means clustering algorithm that is most suitable for the data features of this topic as the follow-up research focus,deduces the process of establishing the Markov chain prediction model in detail,and finally establishes a k-means based Markov chain individual stock trend forecast model based on means of clustering algorithm.Then,taking the stock of China Mobile as an example,an empirical analysis was performed using the established Markov chain prediction model,and the model's prediction accuracy rate was 58.3%.The key to this model is the state division of the Markov chain.The good state division will greatly increase the prediction accuracy of the model.Therefore,this article comprehensively considers several key indicators that affect the stock price,such as the closing price,the ups and downs,and the ups and downs.The state division of the amplitudes solves the problem of Markov State Division and avoids subjective delusion,which is also the innovation point of this article.In the end,based on the aim of continuously improving the accuracy of model prediction,the indicator continues to increase the volume of the status of the indicator and proposes an improved Markov prediction model,which makes the prediction accuracy of the model increase by 27.4%,and finally reach 85.7%.The rationality and referability of the built model can be provided for the general public to provide some suggestions for reference.