Research And Implementation Of Financial Data Analysis System Based On Time Series Data Mining

Financial data is the core of the operation of the financial market.If people need to grasp the development trend of the financial market,they should analyze the main characteristics in the massive financial data and grasp the trend of the financial data in real time.The rapid development of informationization makes the huge and complicated data in the financial background expand again,so the traditional financial analysis method is stretched.Therefore,a financial data analysis system based on time series data mining is designed and implemented to improve the service quality of financial data,improve the management efficiency of financial data,and provide more convenient services for financial data managers.When the financial data analysis system was designed,it was divided into six modules: data collection and update function,financial consulting information management function,financial risk control management function,intelligent mining and analysis function,financial data monitoring function,stock correlation and comparison function.The risk control model of data mining based on time series is established during intelligent mining.The innovation of this model lies in the screening of financial data indexes according to three different levels,among which the first-level index contains the increase and price-to-book ratio,the second-level index contains the stock price and other data of the day,p/e ratio,net interest rate,inflow and outflow,and the third-level index contains assets,gross profit rate and net profit.The method based on time series is used to mine these indexes,and the predicted value of the early data is obtained by calculating the smooth index and the difference index between the data.In order to avoid the possible local influence,the error value was imported into the calculation formula to obtain the specified prediction value.Through the moving average process,the data in the form of linear time series were predicted,and the predicted value was similar to the actual operation value.Since most of the data in ARMA model have the obvious characteristic of periodicity,BP neural network is adopted to further fit the predicted value of irregular financial data.The multi-layer feedforward neural network trained in accordance with the error reverse propagation algorithm outputs the input data in the reverse normalization,and then continuously corrects the internal weights to output the final predicted value according to the sigmoid function.After testing the system,the test results show that the system related functions and performance test results meet the expected requirements.The risk control model of data mining based on time series is tested according to the indexes of different grades,and the results show that the predicted value of this model is up to more than90% under different grades,which can mine financial data.When the system is widely used,it can further provide effective solutions for users,such as the storage of financial data by referring to cloud computing,so as to reduce the load of system operation.