Skip to main content

Table 5 Cluster accuracy on the completely synthetic datasets with different numbers of repeated measurements

From: Clustering gene-expression data with repeated measurements

Number of repeated measurements Noise Similarity measure/mode Average SD-weighted IMM
1 Low Correlation 0.680 NA NA
1 Low Distance 0.789 NA NA
1 Low Spherical NA NA 0.804
1 Low Elliptical NA NA 0.804
1 High Correlation 0.259 NA NA
1 High Distance 0.000 NA NA
1 High Spherical NA NA 0.395
1 High Elliptical NA NA 0.395
4 Low Correlation 0.764 0.576 NA
4 Low Distance 0.877 0.927 NA
4 Low Spherical NA NA 0.926
4 Low Elliptical NA NA 0.957
4 High Correlation 0.389 0.519 NA
4 High Distance 0.000 0.713 NA
4 High Spherical NA NA 0.589
4 High Elliptical NA NA 0.911
20 Low Correlation 0.854 0.701 NA
20 Low Distance 0.891 0.964 NA
20 Low Spherical NA NA 0.962
20 Low Elliptical NA NA 0.957
20 High Correlation 0.602 0.651 NA
20 High Distance 0.590 0.819 NA
20 High Spherical NA NA 0.688
20 High Elliptical NA NA 0.953
  1. Cluster accuracy on the completely synthetic data with different numbers of repeated measurements and different noise levels using average linkage hierarchical clustering algorithm. For each number of repeated measurements and noise level, the highest average adjusted Rand index is shown in bold. As we generated five random synthetic datasets, the results shown are averaged over five synthetic datasets.