点集数据不规则形状时空异常聚类模式挖掘研究

发布时间：2018-07-14 16:10

【摘要】：传统扫描统计方法在进行时空异常聚类模式挖掘时,受扫描窗口形状的限制,不能准确地获取聚类区域形状。提出一种改进的不规则形状时空异常聚类模式挖掘方法stAntScan。新方法基于26方位时空邻近单元格构建时空邻接矩阵,再对蚁群最优化扫描统计方法进行改进,使其能适应三维大数据量的时空区域扫描。模拟数据和真实微博签到数据的实验证明,stAntScan能有效地识别时空范围内的不规则形状异常聚类,并且准确性较经典的SaTScan方法高。
[Abstract]:The traditional scanning statistical method can not accurately obtain the shape of clustering region because of the limitation of scanning window shape when mining spatio-temporal anomaly clustering pattern. An improved clustering pattern mining method for irregular shape spatio-temporal anomalies, stAntScan. is proposed. The new method is based on 26 azimuth spatio-temporal adjacent cells to construct spatio-temporal adjacency matrix, and then improves the ant colony optimization scanning statistical method to adapt to the spatio-temporal scanning of 3D large amount of data. The experimental results of simulated data and real Weibo check-in data show that stAntScan can effectively identify irregular shape anomaly clustering in space-time range, and its accuracy is higher than that of the classical SaTScan method.
【作者单位】：武汉大学资源与环境科学学院;中国科学院亚热带农业生态研究所;
【基金】：国家自然科学基金(41471327,41001231)~~
【分类号】：TP311.13

【参考文献】