A Comparison Between Methods to Cluster Mixed-Type Data: Gaussian Mixtures Versus Gower Distance

In this paper, we compare through a simulation study two approaches to cluster mixed-type data, where some variables are continuous and some others ordinal. The first is model-based, according to which the variables are assumed to follow a Gaussian mixture model, where, as regards the ordinal variables, it is only partially observed. In order to overcome computational issues, the parameter estimation is carried out through an EM-like algorithm maximizing a composite log-likelihood based on low-dimensional margins. In the second approach, the Gower distance matrix is computed, then the PAM algorithm is used for clustering.