Research And Application Of Time Series Data Mining Method For Financial Sector

Data mining technology combines many disciplines such as artificial intelligence,statistics,and stochastic processes,and involves various neural network,wavelet analysis and other model methods.Therefore,data mining technology became the first choice when researching massive data.A time series is an ordered sequence that is preserved in chronological order.Financial timing has dynamic,random,nonlinear,and high noise characteristics compared with general time series data.How to excavate information from these massive data that can guide business behavior has become a hot topic of attention.Due to the particularity of data,traditional data mining methods have certain limitations in clustering and forecasting.To solve this problem,this paper starts with the clustering algorithm and mines the financial sequential data to form a model for prediction.The main work of this paper is as follows:Firstly,it expounds and summarizes the traditional data mining technology.It points out that the current DBSCAN algorithm has the disadvantages of parameter sensitivity and parameter globalize in the clustering process,which makes DBSCAN unable to achieve better data set for changing density in clustering.Clustering effect.Based on this,the method of adaptive parameter selection and initial point optimization of DBSCAN algorithm is proposed,namely OVDBSCAN algorithm.Experiments verify that the OVDBSC AN algorithm has good clustering effect on data sets of different densities.Secondly,based on the OVDBSCAN algorithm,combined with support vector machine regression and parameter optimization techniques,a mixed algorithm(combined-OS algorithm)that combines the advantages of multiple algorithms is designed and implemented.The experimental analysis shows that the algorithm is unstable and non-steady.Effective regression prediction of linear,noisy data.Finally,the mixed-OS algorithm was applied to the stock and financial indexes of the People's Bank of China and China Unicom of the Shanghai Stock Exchange to make predictions to verify the validity of the algorithm.The algorithm is implemented in the system and the financial forecast is presented in the form of a graphical interface.