A Data Mapping Strategy of Parallel Data Mining in Grid

In order to achieve data load balancing of mining nodes in parallel data mining based on grid, this paper proposes a data mapping strategy based on minimum time cost: MapND. Firstly, MapND strategy divided data sets referenced in the data mining into data blocks; Then, the strategy respectively estimated the sum of transmission time and data mining time between each mining node and each data block, and mapped each data block to corresponding mining node according to the minimum time sum. Experimental results show that the proposed MapND strategy can solve the problem of data load balancing for data mining nodes, and improve the performance of parallel data mining in grid.