Improved K-Modes Clustering Method Based on Chi-square Statistics

This paper proposes an improved K-Modes clustering method based on Chi-square statistics, using Chi-square statistics to characterize the relationship between the attributes of data objects. On this basis, the new distance measure is proposed, The distance measure method not only take into account the value of an attribute of an object different from itself, but also take into account other attributes´ influence, which can meet the practical problems. The experimental results show that the proposed clustering method is effective, it can improve the accuracy of the clustering.