PCA - explained variance
Dear mixOmics programmers,
First of all, congratulations for the new version of mixOmics, specially to Dr Le Cao and Dr Rohart. You made great work and can be proud of it. Though I have to point out an issue on the explained variance of the PCA. I do not understand the ouput "explained_variance" of our function. Usually people use eigenvalue/sum(eigenvalue) (see for example FactomineR). In our case you use the square of the eigenvalue. Any reason?
Besides there is a "convergence algorithm issue" in this function. The sum of the eigenvalues should be equal to the dimension of the input data (with ncomp = nrow(input data)). That is not always the case.
Finaly there is a typo in the help file when you explain the parameter logratio with dadta instead of data.
Regards,
B.
Comments (2)
-
-
- changed status to resolved
typo fixed in commit c44b59c
- Log in to comment
Ok I have my answer about the difference between mixOmics and FactominerR. Both are correct. Only the typo need to be fixed. B.