8000 Collapsing PCA and RandomizedPCA · Issue #5243 · scikit-learn/scikit-learn · GitHub
[go: up one dir, main page]

Skip to content

Collapsing PCA and RandomizedPCA #5243

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
giorgiop opened this issue Sep 10, 2015 · 6 comments · Fixed by #5299
Closed

Collapsing PCA and RandomizedPCA #5243

giorgiop opened this issue Sep 10, 2015 · 6 comments · Fixed by #5299

Comments

@giorgiop
Copy link
Contributor

Following discussions in #5141 we may want to collapse RandomizedPCA as an alternative solver into PCA with a svd_solver='auto' that would use randomized_svd whenever n_components + n_oversamples < 0.8 * min(n_samples, n_features).

This reflects the implementation of fbpca. People may not know that RandomizedPCA exists or when it makes sense to use it.

@amueller
Copy link
Member

+1

@kastnerkyle
Copy link
8000 Member

I think it is a good idea - +1

On Thu, Sep 10, 2015 at 9:53 AM, Andreas Mueller notifications@github.com
wrote:

+1


Reply to this email directly or view it on GitHub
#5243 (comment)
.

@jakevdp
Copy link
Member
jakevdp commented Sep 10, 2015

+1 – though of course we'd want to leave RandomizedPCA around as a wrapper for a couple releases.

@larsmans
Copy link
Member

+1

@giorgiop
Copy link
Contributor Author

Any idea on how to handle the options n_components in (0,1) and 'mle'? Those are used after computing the SVD. I also want to start supporting scipy.sparse.linalg.svds as a solver for truncated SVD. The issue is the same.

ValueError or something more clever?

@GaelVaroquaux
984B Copy link
Member
GaelVaroquaux commented Sep 18, 2015 via email

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

6 participants
0