论文部分内容阅读
时空聚类分析是对时空大数据进行利用的一种有效手段。本文提出了一种分布式增量大数据聚类分析方法,利用分布增量机制不但可以减少重复计算和迁移拷贝次数,而且可以持续对聚类结果进行修正,能够在保持聚类准确性的条件下提升整体运算效率。而聚类算法本身通过数据聚集趋势预分析、聚类算法和结果评价3个步骤,构建了一体化时空邻域,在时间和空间维度保证了聚类结果的准确性。经过试验证明该方法可以实现时空大数据的快速高效信息挖掘。
Spatiotemporal clustering analysis is an effective way to use big data in space-time. In this paper, we propose a clustering analysis method for distributed incremental big data, which not only can reduce the number of duplicate computation and migration, but also can continuously correct the clustering results. While keeping the clustering accuracy Under the overall efficiency of the upgrade. The clustering algorithm itself through the data aggregation trend analysis, clustering algorithm and results evaluation of three steps to build an integrated spatio-temporal neighborhood, in time and space dimensions to ensure the accuracy of the clustering results. Experiments show that this method can realize fast and efficient information mining of big data in space-time.