A Data Mining Approach Based on Identical-Different-Contrary Analysis

A new approach,which introduces set pair analysis(SPA) theory,is proposed to analyse the trends of multiple data streams by using the peak data in this paper.The approach simplifies the computation and reduces data volumns.We define two windows,namely: basic window,and observed window.The identical-different—contrary relationship is determined in the former and the set pair trends is counted in the latter,and the strong similar set pair trends pairs are picked out finally.The judgment on the identical-different-contrary relationship and the calculation of the set pair trends are both simple.Moreover,the adoption of the peak data reduces the amount of data greatly,thus the processing time of the data streams is shortened naturally.Experiments have shown that the trends of the peak data can mirror the trends of the data stream vividly.At the same time,the method,employed to solve inconsistency of sample points,is effective.