Skip to main content
Figure 17 | BMC Bioinformatics

Figure 17

From: Dissecting trait heterogeneity: a comparison of three clustering methods applied to genotypic data

Figure 17

Example of k -Modes Clustering. In this example, the same dataset presented in Figure 15 is used to demonstrate the different steps involved the k-modes clustering algorithm. In this example, k was chosen to be 4, such that four clusters will initially be formed. (a) The cluster centroids are seeded by randomly selecting the genotypes of actual individuals in the dataset. The number in parentheses beside the cluster number is the individual used to seed that cluster. (b) Individuals are then compared to each of the cluster centroids, and the number of nonmatching genotypes between each cluster centroid and that individual are recorded. The individual is then assigned to the cluster for which it had the fewest number of nonmatching genotypes (in bold). (c) The next step is to update the cluster centroids based on the individuals now assigned to the clusters. The mode genotype among individuals assigned to a cluster becomes the centroid genotype at that locus. Genotypes that changed from the initialization to the update are shown in bold. (d) After the centroids are updated, the individuals are reevaluated as to which cluster they most closely resemble and are assigned to that cluster. Only cluster 4 was assigned to a different cluster than it was previously. Steps (c) and (d) are repeated until no genotypes are changed in any cluster centroid and no individuals' cluster assignments are changed.

Back to article page