“…For mixed data clustering, in addition to using 1-hot encoding to obtain continuous features or Gower's coefficient [Gower, 1971] and its extensions [Legendre and Legendre, 1998;Podani, 1999] to measure the similarities between data points, as introduced in Section 1, there are also some specially designed clustering algorithms, including k-prototypes [Huang, 1997;], K-means-mixed [Ahmad and Dey, 2007], CAVE [Hsu and Chen, 2007], M-ART [Hsu and Huang, 2008], INTEGRATE [Böhm et al, 2010], INCONCO [Plant and Böhm, 2011], SCENIC [Plant, 2012] and so on. K-prototypes algorithm [Huang, 1997;], which essentially follows the same idea of k-means algorithm, calculates the dissimilarity between two mixed-type objects as a combination of the squared Euclidean distance measure on the numeric attributes and the simple matching dissimilarity measure on the categorical attributes.…”