In principal coordinates analysis with the presence of negative eigenvalues, what's the best way to calculate the percentage of variation explained by each principal coordinate? Does it make sense to consider only the principal coordinates corresponding to the positive eigenvalues, and calculate the percentage of variation explained by each principal coordinate by dividing the positive eigenvalue by the sum of the positive eigenvalues (i.e. discarding the negative eigenvalues)?
Asked
Active
Viewed 758 times
2
-
There was almost a duplicate question already: http://stats.stackexchange.com/q/99009/3277. Despite it has not been answered formally, comments there share some of my thoughts. – ttnphns May 16 '16 at 08:02