Application Research on time series data mining outliers in linear programming

Abstract

Based on data mining of outliers of time series, current studies have developed many methods, but there exist certain disadvantages for each method. This paper developed the method of mining outliers of time series based on linear programming model, verified high efficiency through an actual case and make up for the shortcoming of other methods. At present, in many scientific fields have accumulated large amounts of data. Data often contains some unknown information. In such circumstances gradually developed a new discipline that is data mining. the data mining has some data with time obviously. this is the time series data mining. In time series data mining work often people ignore the abnormal data. However, some abnormal data is able to respond to some unexpected situations. Especially in the electronic commerce environment. the time series data mining technology can help the electronic commerce enterprise knowledge from data mining, potential rules. Then these findings shown by visualization method. It can help enterprises to deepen the understanding of knowledge acquisition. The optimization of enterprise decision management, customer relationship management, collaborative business management, marketing management, website maintenance management and risk control management. confirm the target market. the realization of personalized marketing, greater competitive advantage. Method of time series outlier data mining has many. Such as: The statistical method, distance based method, density based method, biological method, based on model method etc. But each method all have different limitations. in this paper based on time series outlier data mining is a mining method of linear programming based on time series outlier data. And a good effect is obtained by the empirical analysis of the 2011 stock data. Hope this method can increase the accuracy of time series the anomaly detection.