A Complete Attribute Reduction Algorithm Based on Improved FP Tree

There are lots of repeat and unnecessary elements in discernibility matrix, which affect attribute reduction algorithm based on discernibility matrix. To improve the efficiency of such algorithms, a novel data structure IFP(improved frequent pattern) tree is proposed, which combine with the idea of FP tree and then can get rid of all the repeat and unnecessary elements in the discernibility matrix. Then, a new complete attribute reduction algorithm is designed based on IFP_Tree. The new algorithm can not only reduce a great deal of memory space, but also enhance the efficiency of attribute reduction algorithm greatly. The theoretical analysis and experimental results show that the new algorithm is more efficient than the existing attribute reduction algorithm based on discernibility matrix, and more adaptive for mining very large datasets.