suggest_best_k-ConsensusPartition-method {cola} | R Documentation |
Suggest the best number of partitions
## S4 method for signature 'ConsensusPartition' suggest_best_k(object, jaccard_index_cutoff = 0.95)
object |
A |
jaccard_index_cutoff |
The cutoff for Jaccard index compared to previous k. |
The best k is selected according to following rules:
1. k with rand index larger than jaccard_index_cutoff
are removed. If all k are removed, the best k is defined as NA
.
2. If there are some k having 1-PAC
larger than 0.9, the largest k is selected as the best k.
3. If it does not fit rule 2, the k with highest vote of highest 1-PAC, mean_silhouette and concordance scores is
selected as the best k.
suggest_best_k
function only gives suggestion of selecting
a reasonable best k. Users still need to look at the plots (e.g. by select_partition_number
or consensus_heatmap
functions), or even
by checking whether the subgrouping gives a reasonable signatures by get_signatures
, to pick a reasonable k that best explains their study.
The best k with 1-PAC larger than 0.9 is treated as a stable partition.
The best k.
Zuguang Gu <z.gu@dkfz.de>
data(cola_rl) obj = cola_rl["sd", "kmeans"] suggest_best_k(obj)