nsh87 / receptormarker

Source for 'receptormarker' package for R: antibody receptor and phenotypic marker analysis
http://receptormarker.com
BSD 2-Clause "Simplified" License
4 stars 9 forks source link

Optimal number of cluster not correct on wss_plot and gap_plot #25

Open nsh87 opened 9 years ago

nsh87 commented 9 years ago

They are both reporting optimal number of clusters as two when using the iris dataset, but this dataset is known to have three clusters.

WSS: screen shot 2015-07-14 at 8 49 06 am

Gap: screen shot 2015-07-14 at 8 48 55 am

nsh87 commented 9 years ago

Same with avg_sil_plot(): screen shot 2015-07-14 at 8 58 58 am

nsh87 commented 9 years ago

This usage of NbClust in multi_clust that was supposed fix this has been reverted in https://github.com/nsh87/receptormarker/commit/46d3e4ed29d15d9cc2d45e40c9ad18a3866da6d9 because NbClust was failing.