Research And Implementation Of Multidimensional Time-Series Data Mining Methods

Data mining is one of hotspots of computer science. Time-Series mining is an important part of data mining research. Compared with other mature parts (such as association mining) of data mining, time-series mining is a newer direction. Especially, multidimensional time-series mining is a much newer direction. Today, the research of time-series mining is gradually becoming a new hotspot. The purpose of time-series mining is to find frequent subsequences, trends, and patterns. Multidimensional time-series mining needs to reduce dimensions.In this paper, the main work is to research the algorithms of dimensionality reduction, similarity search and visualization. The improvement or the optimization is made for each algorithm. Moreover, I demonstrate them through experiments, and give the related results. The platform based on web service, which is used for visual data mining, is implemented.The main research contents are summarized as follows:1) Dimensionality Reduction: The dimensionality reduction can overcome the curse of dimensionality and model high dimensional data.