| With the development and popularization of data processing technology,the accumulatio n of human's data and statistics is rapidly expanding.However,the true value of these data ha s not been fully exploited and employed,which gives a platform of development for Data Min ing.There are large amount of time series data in finance data.Researchers currently emphasize how to acquire valuable and effective information from time series data.However,the difficulty and also a heated research area lie in the Prominent Streak,which is of considerable theoretical and practical significance.To detect Prominent Streak in time series data is conducive to the discovery of questionable experiment phenomenon in multiple tests and also to the precise analysis of industry trend,especially in banking and securities.This paper studies the application of a significant financial sector in a continuous sequence.Timing data having a high frequency,non-stationary,pseudo-cyclical characteristics,based on existing financial industry data mining model,to fully understand its advantages and shortcomings,we propose a significant time-series method to applicate in high frequency data of the financial industry,by introducing the concepts and methods of the discrete Fourier transform in this idea,we propose a significant based on pseudo-periodic sequence of continuous mining method.Basised on given the dynamic pseudo-periodic intervals and space pruning ideas,we achieve a method,ISLPS,what can mining PLPS effective,proved ISLPS effective in prominernt streak in lab,so that the significant time series in the financial sector and It has good application.Finally,we will propose a significant continuous sequence mining algorithm(ISLPS)ap plied to large data analysis platform,securities index system,results indicates a significant co ntinuous sequence mining algorithm in actual production,which generated proved remarkably researching prominent sequence series mining and its meaning and value. |
PDF Full Text Request