k-ANMI: A mutual information based clustering algorithm for categorical data

Abstract

Clustering categorical data is an integral part of data mining and has attracted much attention recently. In this paper, we present k-ANMI, a new efficient algorithm for clustering categorical data. The k-ANMI algorithm works in a way that is similar to the popular kmeans algorithm, and the goodness of clustering in each step is evaluated using a mutual information based criterion (namely, average normalized mutual information -ANMI) borrowed from cluster ensemble. This algorithm is easy to implement, requiring multiple hash tables as the only major data structure. Experimental results on real datasets show that k-ANMI algorithm is competitive with those state-of-the-art categorical data clustering algorithms with respect to clustering accuracy.

Spectral clustering and semi-supervised learning using evolving similarity graphs

“…Some of the most widely used internal criteria are Calinski-Harabasz index [28], Davies-Bouldin index [29] and Dunn's index [30], whereas some external criteria are F-measure [31], purity [32], normalised mutual information [33] and a measure based on hungarian algorithm [34]. All the aforementioned criteria have been used in the proposed algorithm, some of them both for optimisation and evaluating the performance of the algorithm and some only for evaluation.…”

“…• Normalised mutual information [33] is also used as an external criterion for evaluating clustering algorithms, and is defined as:…”

EEG emotion recognition using reduced channel wavelet entropy and average wavelet coefficient features with normal Mutual Information method

“…However, its implementation has been applied in data mining to group similar data records [8]. Recently, Normalized Mutual Information (NMI) has been proposed for feature selection with the advantage of reduced complexity of features [9].…”

Clustering Categorical Data Based on Within-Cluster Relative Mean Difference

“…The issue of database clustering with categorical variables has received intensive attention ( [1]- [8]) along with other publications in the same issue. The categorical data come from different areas of research, both social and nature sciences; this type of variable does not present a natural ordering for their possible values, results in the difficulty in clustering process.…”