[MRG] Updated partial_fit documentation #12618
Conversation
| """Fit the model to data matrix X and target y. | ||
| """Fit the model to data matrix X and target y for a single iteration. | ||
|
|
||
| Reuses the previously fitted solution as initialization if is exists. |
There was a problem hiding this comment.
I think using initialization is confusing here. That would be warm-start.
Maybe "do a single optimization iteration over the data, starting from the previous state"?
There was a problem hiding this comment.
That's fine with me, this new commit uses that language. (Although under-the-hood they technically work the same)
There was a problem hiding this comment.
They do not, see the glossary. partial_fit does a single iteration, warmstart solves till the optimum.
There was a problem hiding this comment.
Yes I understand. I just meant that in the context of the MLP classes, the way partial_fit and warm_start pick up from the previous stopping point is the same. But I agree using the word "initialization" is more appropriate for warm start than partial.
| @@ -619,7 +619,8 @@ def fit(self, X, y): | |||
|
|
|||
| @property | |||
| def partial_fit(self): | |||
| """Fit the model to data matrix X and target y. | |||
| """Fit the model to data matrix X and target y for a single | |||
There was a problem hiding this comment.
How about "update the model with a single iteration over the given data"
This would also satisfy pep257's request that we keep this summary to a single line
There was a problem hiding this comment.
I like it, done.
This reverts commit 457d2d7.
This reverts commit 457d2d7.
Reference Issues/PRs
Addresses #12505 documentation
What does this implement/fix? Explain your changes.
Updates the documentation to make the behavior of
partial_fitmore clear. I think this is the last change from #12505 to be made.