suggest_best_k-HierarchicalPartition-method.RdGuess the best number of partitions
# S4 method for HierarchicalPartition
suggest_best_k(object)A HierarchicalPartition-class object.
It basically gives the best k at each node.
A data frame with the best k and other statistics for each node.
data(golub_cola_rh)
suggest_best_k(golub_cola_rh)
#> node best_method is_leaf best_k 1-PAC mean_silhouette concordance n_sample
#> 0 ATC:skmeans 3 1.000 0.964 0.984 72 **
#> 01 ATC:skmeans 3 1.000 0.951 0.982 35 **
#> 011 not applied ✓(b) NA NA NA NA 11
#> 012 ATC:skmeans ✓(c) 3 1.000 0.987 0.994 13 **
#> 013 not applied ✓(b) NA NA NA NA 11
#> 02 ATC:skmeans 4 0.913 0.901 0.944 24 *
#> 021 ATC:skmeans ✓(a) 2 0.846 0.898 0.965 13
#> 022 not applied ✓(b) NA NA NA NA 11
#> 03 ATC:skmeans ✓(c) 2 0.705 0.932 0.971 13
#> ---------------------------------------------------------------------------
#> Stop reason:
#> a) Mean silhouette score was too small
#> b) Subgroup had too few columns.
#> c) There were too few signatures.