关键词: 数据预处理, 缺失数据, 最近邻算法, 灰色分析, 马氏距离

Abstract: The Euclidean-based k-Nearest Neighbor (kNN) algorithm is restricted to the dataset without correlation-sensitive on density. The author proposed an improved kNN algorithm based on Mahalanobis distance and gray analysis for imputing missing data to replace the existing Euclidean distance. The Mahalanobis distances can deal with the issue of correlation-sensitive on density, and the gray-analysis method can deal with the opposite case. Hence, the proposed method can deal with any kind of datasets, and the experimental results show the proposed method outperforms the existing algorithms.

Key words: data preprocessing, missing data, Nearest Neighbor (NN) algorithm, gray analysis, Mahalanobis distance