Skip to main content

Table 1 Summary of various clustering approaches used in our empirical study

From: Clustering gene-expression data with repeated measurements

Clustering algorithms

Similarity measures

Approach to repeated data

Hierarchical agglomerative (average linkage, centroid linkage, complete linkage, single linkage)

Correlation/distance

Average over repeated measurements variability-weighted similarity. Force into the same subtree (FITSS)*

k-means

Correlation/distance

Average over repeated measurements variability-weighted similarity

CAST

Correlation/distance

Average over repeated measurements variability-weighted similarity

DIANA (hierarchical divisive)

Correlation/distance

Average over repeated measurements variability-weighted similarity

MCLUST-HC

None

Average over repeated measurements. Force into the same subtree (FITSS)*

IMM

None

Built-in error models (spherical, elliptical)

  1. *FITSS refers to clustering the repeated measurements as individual objects and force the repeated measurements into the same subtrees. MCLUST-HC denotes a model-based hierarchical clustering algorithm as implemented in the hcVVV function in the 2002 version of MCLUST.