* FIX test_common: assert_greater_equal(iter_, 1)
The previous check was too strict. It is fine for an estimator to
decide that is has converged in one pass over the data.
* STYLE PEP8
* FIX test_reconstruct_patches_perfect_color too strict
* STYLE flake8 fixes
Ensuring consistent transforms for KernelPCA
Taking @vene's changes into account, thanks!
Taking @jakevdp's comment into account
Added more verbose documentation to kernel_pca.py
Specifying that X_fit_ will not be None. @jakevdp
KernelPCA: Fixing more formatting of docstring
Addressed @vene's documentation comments
Addressing that dual_coef_ might not be present in model in docs. @vene
Added the following things:
* Test when the sum is not equal to one
* Test the prediction in case of a large bias for one class
* Test explicitely class_prior_
* Move the function to update the class prior in the GaussianNB class
* Remove the updating of the class prior before to actually compute the mean and variance
Address comments for PR #6180 - Correct the documentation
Address comment PR#6180 - Improve the class prior initialisation and updating
We modify the code to:
* Initialisat self.class_prior_ with the different possibilities (class_prior given or not, fit_prior True or False)
* Update self.class_prior_ only when no class_prior is given and than fit_prior is True
Address comments PR #6180 - Remove useless line
Fix the file according to PEP8 regulations
Update the API to have only class_prior in GaussianNB
Correct prior fitting using samples and not class number
Change name of priors and correct the warning with division by zero
Update the API
Remove functions which were called only once
Add additional test for part of the code which was not covered in GaussianNB
Correct doc formatting
Correct spelling
- total_n_estimators should be sum(n_estimators) not len(..)
- keyword argument should be after non-keyword argument (call of ``_parallel_build_estimators`` with ``delayed`` on line 365-366.
Add to bagging.Bagging._parallel_build_estimators a total_n_estimators parameter, to improve the (optional) information printed if verbose > 1.
It felt weird to see 6 times in a row "building estimator 1 of 1" without any idea of total number of estimators to build, now it's fixed.
Register OneVsRestRegressor as meta estimator
Rename to a more sensible name
Parallel predict and sparse support
Started MultiOutput documentation
Move code to new file multioutput.py
Continuing the move to new multioutput module
Added sample weight support
Better test for sample weights and actually support weights
Added a new test using weighted vs repeated samples to
test sample weight support. Uncovered that weights
were not actually passed on to underlying estimator.
Comment on multiprocess overheads
Move parallel_helper to utils.fixes
This helper works around a python2 limitation on pickling
instance methods
Example of multi-output regression with gradient boosting
Switch to uniform weighted score and updated example
The example now uses a RF with and without the MultiOutput
meta estimator
Added note for removing `score` method
Addressing comments on MultiOutputRegressor
MultiOutputregressor better test for weighted samples
Fix ups
Use explicit keyword argument for passing sample weights and
fix random_state on train-test split in the example
Olivier Grisel