search

matthias-da / robCompositions

Robust Methods for Compositional Data
11 stars 8 forks source link

pcaCODA doesn't work with microbiome data #10

Closed Jayanth-kumar5566 closed 3 years ago

Jayanth-kumar5566 commented 3 years ago

microbiome dataset usually has a higher number of features than the number of samples. This is causing a problem in the calculation of covariance using MCD.

``

expenditures<-expenditures[1:4,] p1 <- pcaCoDa(expenditures) Error in robustbase::covMcd(xilr, cor = FALSE) : n <= p -- you can't be serious! ``

alexkowa commented 3 years ago

I don't see a possibility to easily extend this to the case of p>n

Jayanth-kumar5566 commented 3 years ago

I thought so, too. Anyways, I will just implement classical PCA on the ILR coordinates and then transform back.