Linear Discriminant Analysis for Starters | Eigenfoo
utterances-bot opened this issue · 4 comments
utterances-bot commented
Linear Discriminant Analysis for Starters | Eigenfoo
Everything that you wanted to know (and more!) about linear discriminant analysis (LDA) — how it works, why it works, and how to use it.
skroon commented
Hi - thanks for the post!
Just two things I picked up here:
- you mixed up your notation S_W and S_B when introducing the within- and between-class scatter matrices (e.g. "define SB to be the within-class scatter matrix").
- the within class scatter matrix is obtained by adding the individual class scatter matrices, not covariance matrices - i.e. your definition of \Sigma_i should be the class scatter matrix, not the covariance matrix, to prevent the classes being equally weighted.
eigenfoo commented
Alxmrphi commented
George, the mixup between S_W and S_B is still in the post. The italicised definitions are the wrong way around. You said in the other comment it was fixed, but it doesn't look it. I think maybe you changed the symbols but not the part that came after that. It could easily lead to confusion so would be nice if you can fix that :)