M Learning / PCA / Use n_components=None by default
When using PCA, the current default number of components is equal to 5 without any good reason. We should use n_components == min(n_samples, n_features)
with the argument value None
, as it is the case with the interfaced PCA from sklearn. Same should be applied for KPCA
, PLS
and KLSVD
Edited by Matthias De Lozzo