Research On The Model Of Context-based Spatio-Temporal Outlier Detection

Context-based Spatio-Temporal outliers are the patterns which are unknown and potentially useful important information. It shows the specificity process of geographical phenomena or geographic. Spatio-Temporal outlier detection plays an important role in several application domains such as animal and plant ecology, geological disaster monitoring, environmental protection, geophysical exploration, public safety and health fields. Recently, background information based on temporal changes in the outlier detection problem is more and more interesting, namely context-based Spatio-Temporal outlier detection. In this paper, we study the context-based Spatio-Temporal outliers, and propose a model of context-based Spatio-Temporal outliers, and give an effective algorithm and the improved method. This paper includes:(1) The introduction of spatial outlier detection data mining and development is described. We propose the detailed discussion of the contextual outlier detection, and research for context-based Spatio-Temporal outliers.(2) Suppose a detail definition of the context-based Spatio-Temporal outliers. A framework of context-based Spatio-Temporal outlier detection is described. We complete the modeling of context-based Spatio-Temporal outliers and analyze completeness, correctness, and time complexity.(3) Give a detail introduction on the specific steps of context-based Spatio-Temporal outlier detection. An effective algorithm of the context-based Spatio-Temporal outlier detection is proposed using a random walk graph methods and other.(4) Improve the proposed algorithm of context-based Spatio-Temporal outlier detection. Based on the feature that outliers are the minority consisting of fewer instances, by adding thresholds and other methods, we reduce the data for most normal operations and optimize the algorithm. It shows the results is efficient and validate for application using real dataset.