If I try and write PCA from memory in PyTorch I always make a mistake so it doesn't do exactly the same thing as scikit-learn's PCA with the same settings. This is a minimal implementation of PCA that matches scikit-learn's with default settings (run pca.py to test this).

Now includes:

• pca.py: matches sklearn's PCA
• incremental_pca.py: matches sklearn's IncrementalPCA, contributed by: yry

Install

Open pca.py, copy the code you'd like to use and then paste it where you'd like to use it.

Related Work

• valentingol's torch_pca appears to be a more full featured and faster (it chooses an appropriate PCA algorithm depending on input dimensions) alternative PCA implementation also matching scikit-learn.