From b801aab95dd9e0610df08db5c256879047ddfa31 Mon Sep 17 00:00:00 2001 From: Yuchen Pei Date: Fri, 4 Jan 2019 15:12:46 +0100 Subject: added a remark about scikit-learn --- posts/2019-01-03-discriminant-analysis.md | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/posts/2019-01-03-discriminant-analysis.md b/posts/2019-01-03-discriminant-analysis.md index 3e732db..51630b3 100644 --- a/posts/2019-01-03-discriminant-analysis.md +++ b/posts/2019-01-03-discriminant-analysis.md @@ -176,6 +176,11 @@ will result in a lossy compression / regularisation equivalent to doing analysis](https://en.wikipedia.org/wiki/Principal_component_analysis) on $(M - \bar x) V_x D_x^{-1}$. +Note that as of 2019-01-04, in the scikit-learn implementation, the prediction is done without +any lossy compression, even if the parameter `n_components` is set to be smaller +than dimension of the affine space spanned by the centroids. +In other words, the prediction does not change regardless of `n_components`. + ### Fisher discriminant analysis The Fisher discriminant analysis involves finding an $n$-dimensional -- cgit v1.2.3