A comparison of latent class, K-means, and K-median methods for clustering dichotomous data.

Abstract

The problem of partitioning a collection of objects based on their measurements on a set of dichotomous variables is a well-established problem in psychological research, with applications including clinical diagnosis, educational testing, cognitive categorization, and choice analysis. Latent class analysis and K-means clustering are popular methods for partitioning objects based on dichotomous measures in the psychological literature. The K-median clustering method has recently been touted as a potentially useful tool for psychological data and might be preferable to its close neighbor, K-means, when the variable measures are dichotomous. We conducted simulation-based comparisons of the latent class, K-means, and K-median approaches for partitioning dichotomous data. Although all three methods proved capable of recovering cluster structure, K-median clustering yielded the best average performance, followed closely by latent class analysis. We also report results for the three methods within the context of an application to transitive reasoning data, where it was found that the three approaches can exhibit profound differences when applied to real data.

Idiopathic Short Stature Phenotypes among Korean Children: Cluster Analysis

“…We switched from a "variable-centered" to a "personcentered" approach to understanding the characteristics of ISS. Clustering methods are useful for grouping subjects into profiles that summarize shared aspects of a disease within a heterogeneous group (Brusco et al 2017). It has already been used in studies of respiratory diseases (Haldar et al 2008;Herr et al 2012;Dumas et al 2016), but has never been applied to children with ISS.…”

Affinity propagation: An exemplar‐based tool for clustering in psychological research

“…However, these methods are less effective, or even inapplicable, for certain types of data. For example, Brusco, Shireman, and Steinley () recently found that K ‐median clustering was better than K ‐means at recovering true cluster structure when objects are measured on binary variables. In light of the similarities (discussed in Section 2.3) between K ‐median clustering and affinity propagation, the latter method is also likely to perform well for such data.…”

Development of technique for face detection in image based on binarization, scaling and segmentation methods

“…Traditional methods of the taxonomic approach are: 1) methods based on partitioning (partition-based, partitioning-based) or center (center-based) (for example, k-means [15], PAM (k-medoids) [16], FCM [17], ISODA-TA [18]);…”