[MRG] Improve speed of plot_pca_vs_fa_model_selection.py example #13685
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
agramfort
reviewed
Apr 21, 2019
|
|
||
|
|
||
| def compute_scores(X): | ||
| pca = PCA(svd_solver='full') | ||
| fa = FactorAnalysis() | ||
| fa = FactorAnalysis(tol=1) |
There was a problem hiding this comment.
I am not sure it's a good idea to advertise using such a high tolerance without a comment or empirical evidence that it's ok.
my 2c
There was a problem hiding this comment.
@agramfort Thanks for your comment. I totally agree! Do you think that the comment introduced in d79e45d is enough?
I also added the same comment in b742f21 for the plot_rbm_logistic_classification.py example where I also increased the tolerance in #13648
|
please show us how the tol affects the results. For FA how many iterations
do you end up doing with tol=1 vs default?
|
|
I'm closing this one as the issue has been solved in #21671. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Reference Issues/PRs
Partial fix to #13383
What does this implement/fix? Explain your changes.
I propose two modifications to the example
examples/decomposition/plot_pca_vs_fa_model_selection.pymentioned in #13383 :FactorAnalysis(tol=1)n_components = 45.The build time (on my computer) decreases from 11.846 to 3.437 seconds.
These modifications are motivated by the fact that among the 3 dimensional reduction techniques trained, only Factor Analysis for 40 and 45 components takes some significant time to fit:
After applying the 2 modifications mentioned above, we achieve significant speed improvements:
The printed chosen best numbers of components remain unchanged (I added the nature of the noise to improve its readability):
However the 2 plots changed a bit:
Homoscedastic Noise
Original plot
Modified plot
Heteroscedastic Noise
Original plot
Modified plot
As you can see, estimated FA scores (average log-likelihood) are different for 40 components for both noises (and the points for 45 components are removed from the 2 plots). However I think that the plots are ok, because the interpretation is left unchanged as this part of the plot isn't important in the example.