Clustering Categorical Data Using Community Detection Techniques

Algorithm 1

CD-Clustering.

Input: dataset with data points, each data point has

attributes. The number of clusters .

Output: clusters of data points .

compute pairwise Hamming distances

compute CDF for pairwise Hamming distances

// estimate distance threshold

for     do

(5) if   and   then

(6)break

(7) // clustering

(8)

(9)for     do

(10) if     then

(11)

(12) run Louvain method [12] on

(13) keep top- communities by size

(14)for each cluster   do

(15) compute the mode of

(16)for each remaining data point   do

(17) assign to the nearest mode ,

(18)return