基于MapReduce的Canopy-Kmeans算法的并行化

Parallelized Canopy-kmeans algorithm based on MapReduce

数据挖掘的聚类算法Canopy-Kmeans是分析数据内在价值的常用工具之一,传统的基于集中控制的方式算法执行效率,在今天大数据环境下,有待改进。文章数据源为某省运营商在2014年7月经过脱敏后的话单信令数据,通过传统的集中控制方式和基于MapReduce的方式。通过实验,我们可以看出使用MapReduce方式具有良好的可行性,而且执行效率也得到明显改善[1]。

The Canopy - kmeans clustering algorithm for data mining is one of the common tools which we usually used to analyze the intrinsic value of data. Under current big data environment, the traditional algorithm based on centralized control need to be improved. In this paper, the data source is gathered July 2014 from the desensitized signal data, and billed by traditional centralized control and the method based on MapReduce. Through the experiment we know it has good feasibility to use the MapReduce way, and the executive efficiency has been improved.