An Approach for Treatment of the Incomplete Data Based on WaveCluster and Weighted 1-Nearest Neighbor

Abstract

For the incomplete data that usually exists in the process of pretreatment, this article presents an approach for treatment of the incomplete data based on WaveCluster and weighted 1-Nearest Neighbor (1-NN).The proposed method firstly carries out the WaveCluster in the complete record set of the whole set, which can reduce the volume of comparative data and rule out outliers, improve computational efficiency of the algorithm and the clustering accuracy. Then, the weighted 1-NN method is used, according to the contribution attributes made to the classification in the algorithm, the information gain of attribute is calculated and each attribute is endowed with certain weight using in the nearest neighbor measure, thus it can enhance the filling precision of the missing value. Experimental results show the proposed method is an appropriate and effective method in treatment of the incomplete data.