Research And System Implementation Of Large Data Mining Based On Trend Analysis Of Financial Price

As the rapid development of the economy,uncertainty of financial activities has been increasing dally.The academia and the financial industry has paid great attention to it,on account of people design to obtain a predictive judgment and speculation,grasp the prospective development tendency and regularity of the financial world.Recently,with the rapid development of the Internet,massive financial information relying on the price history of the data mining technology solely can't reflect the impact of the multiple factors in financial market.Therefore,the paper proposes the method digging the emotional tendency messages from the financial news information,combining with financial historical price data,using multiple linear regression method on the financial market of the future development trend of the prices prediction to obtain more accurate prediction results.In the process of forecasting,the multivariate linear model is used for preliminary processing,and then the residual model is corrected.And use Spark technology to optimize the performance prediction process,and achieved good results.The main research contents of this paper include the following:(1)The method is proposed to quantify the financial news of emotion information.Convert the collected news as vectors,and then through the moral sentiment lexicon and custom rules,combined with the history financial price,we get every financial news articles emotional value,so as to obtain for a kind of emotional situation of financial products every day.The financial news and financial data with the historical price prediction based on the proposed use of multiple linear regression prediction,and then use a combination of model error correction autoregressive moving average model prediction,the predicted results are satisfactory.(2)This paper proposes a method to quickly obtain the emotion value of financial markets using Spark Streaming technology.In order to improve the prediction performance,the emotion value acquisition process under the Spark environment can get better speed performance,because of the large amount of financial information and data,the complex process and the process of obtaining the information of emotion preference value.(3)Based on the combination of textual information of financial and financial information and the prediction method of financial market historical data,this paper constructs the forecasting system of financial market development trend.Show the results of this method of forecasting at the same time,better for the majority of investors to create a guide investment platform.