Refactor StackingCVRegressor so that users can obtain meta_features #294

takashioya · 2017-11-30T03:03:56Z

Description

modify fit and add an attribute self.train_meta_features_ for obtaining train meta-features,
and add a method predict_meta_features for obtaining test meta-features.

Related issues or pull requests

#285
#290

Pull Request requirements

Added appropriate unit test functions in the ./mlxtend/*/tests directories
Ran nosetests ./mlxtend -sv and make sure that all unit tests pass
Checked the test coverage by running nosetests ./mlxtend --with-coverage
Checked for style issues by running flake8 ./mlxtend
Added a note about the modification or contribution to the ./docs/sources/CHANGELOG.md file
Modify documentation in the appropriate location under mlxtend/docs/sources/ (optional)
Checked that the Travis-CI build passed at https://travis-ci.org/rasbt/mlxtend

pep8speaks · 2017-11-30T04:16:10Z

Hello @takashioya! Thanks for updating the PR.

In the file mlxtend/regressor/tests/test_cv_stacking_regression.py, following are the PEP8 issues :

Line 124:42: E231 missing whitespace after ','

Comment last updated on December 01, 2017 at 07:02 Hours UTC

coveralls · 2017-11-30T05:36:43Z

Coverage increased (+0.02%) to 90.897% when pulling fc4f3ef on takashioya:refactor-meta-features into cca77ac on rasbt:master.

coveralls · 2017-11-30T08:06:08Z

Coverage increased (+0.02%) to 90.897% when pulling c77edb3 on takashioya:refactor-meta-features into cca77ac on rasbt:master.

takashioya · 2017-11-30T08:13:57Z

Sadly, flake8 ./mlxtend still gives many errors in my environment. I think, however, most of them are not related to my commit.

coveralls · 2017-11-30T08:23:09Z

Coverage increased (+0.02%) to 90.897% when pulling 2762a64 on takashioya:refactor-meta-features into cca77ac on rasbt:master.

coveralls · 2017-11-30T08:46:47Z

Coverage increased (+0.02%) to 90.897% when pulling d5b6f73 on takashioya:refactor-meta-features into cca77ac on rasbt:master.

rasbt · 2017-11-30T18:02:40Z

Sadly, flake8 ./mlxtend still gives many errors in my environment. I think, however, most of them are not related to my commit.

No worries, you can ignore those (and I can fix them another day). Usually, I say that making sure that any new code is "PEP8"-fine suffices :).

Thanks for the PR, I will take a look at it now and provide my comments :)

rasbt

The PR looks great so far. Here are just some minor suggestions

rasbt · 2017-11-30T18:04:27Z

docs/sources/CHANGELOG.md

@@ -15,6 +15,8 @@ The CHANGELOG for the current development version is available at
 ##### New Features

 - New `max_len` parameter for the frequent itemset generation via the `apriori` function to allow for early stopping. ([#270](https://github.com/rasbt/mlxtend/pull/270))
+- New `store_train_meta_features` parameter for `fit` in StackingCVRegressor. if True, train meta-features are stored in `self.train_meta_features_`.
+    New `pred_meta_features` method for StackingCVRegressor. People can get test meta-features using this method. ([#294](https://github.com/rasbt/mlxtend/pull/294))


Unless you really don't like to be mentioned, you can add your name or GitHub handle here. E.g.,

New `pred_meta_features` method for StackingCVRegressor. People can get test meta-features using this method. ([#294](https://github.com/rasbt/mlxtend/pull/294) via [takashioya](https://github.com/takashioya))

rasbt · 2017-11-30T18:09:48Z

mlxtend/regressor/stacking_cv_regression.py

@@ -66,10 +66,14 @@ class StackingCVRegressor(BaseEstimator, RegressorMixin, TransformerMixin):
        be shuffled at fitting stage prior to cross-validation. If the `cv`
        argument is a specific cross validation technique, this argument is
        omitted.
+    store_train_meta_features : bool (default: False)
+        If True, meta-features for training data is stored when you `fit`
+        training data. You can get them by `self.train_meta_features_`.


I suggest a minor rephrasing to

store_train_meta_features : bool (default: False) If True, the meta-features computed from the training data used for fitting the meta-regressor stored in the `self.train_meta_features_` array, which can be accessed after calling `fit`.

Could you also describe the shape/layout of the self.train_meta_features_ array?

rasbt · 2017-11-30T18:12:16Z

mlxtend/regressor/stacking_cv_regression.py

@@ -163,6 +172,14 @@ def predict(self, X):
        else:
            return self.meta_regr_.predict(meta_features)

+    def predict_meta_features(self, X):


Here, it would be helpful to format it as a docstring so that it shows up in the API documentation. E.g., something similar to the fit docstring. I just see that predict, get_params etc. don't have docstrings as well, but I can fix that later separately after this PR.

coveralls · 2017-12-01T07:13:30Z

Coverage increased (+0.02%) to 90.897% when pulling 3067f26 on takashioya:refactor-meta-features into cca77ac on rasbt:master.

coveralls · 2017-12-01T07:28:55Z

Coverage increased (+0.04%) to 90.916% when pulling 61861dc on takashioya:refactor-meta-features into cca77ac on rasbt:master.

coveralls · 2017-12-01T07:29:15Z

Coverage increased (+0.04%) to 90.916% when pulling 61861dc on takashioya:refactor-meta-features into cca77ac on rasbt:master.

coveralls · 2017-12-01T07:33:33Z

Coverage increased (+0.02%) to 90.897% when pulling 154499e on takashioya:refactor-meta-features into cca77ac on rasbt:master.

takashioya force-pushed the refactor-meta-features branch from 2762a64 to d5b6f73 Compare November 30, 2017 08:28

rasbt mentioned this pull request Nov 30, 2017

Some minor typo fixes in the StackingCVRegressor docs #295

Closed

rasbt requested changes Nov 30, 2017

View reviewed changes

takashioya force-pushed the refactor-meta-features branch from dc0a15b to 61861dc Compare December 1, 2017 07:21

takashioya closed this Dec 1, 2017

takashioya force-pushed the refactor-meta-features branch from 61861dc to 154499e Compare December 1, 2017 07:31

takashioya mentioned this pull request Dec 1, 2017

Refactor StackingCVRegressor so that users can obtain meta_features #296

Merged

7 tasks

This was referenced Dec 4, 2017

Attribute and method store/compute meta features of StackingRegressor #299

Merged

Add meta feature functionality to StackingClassifier and StackingCVCl… #300

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Refactor StackingCVRegressor so that users can obtain meta_features #294

Refactor StackingCVRegressor so that users can obtain meta_features #294

takashioya commented Nov 30, 2017 •

edited

Loading

pep8speaks commented Nov 30, 2017 •

edited

Loading

coveralls commented Nov 30, 2017

coveralls commented Nov 30, 2017

takashioya commented Nov 30, 2017

coveralls commented Nov 30, 2017

coveralls commented Nov 30, 2017

rasbt commented Nov 30, 2017

rasbt left a comment

rasbt Nov 30, 2017

rasbt Nov 30, 2017

rasbt Nov 30, 2017

coveralls commented Dec 1, 2017

coveralls commented Dec 1, 2017

coveralls commented Dec 1, 2017

coveralls commented Dec 1, 2017

Refactor StackingCVRegressor so that users can obtain meta_features #294

Refactor StackingCVRegressor so that users can obtain meta_features #294

Conversation

takashioya commented Nov 30, 2017 • edited Loading

Description

Related issues or pull requests

Pull Request requirements

pep8speaks commented Nov 30, 2017 • edited Loading

Comment last updated on December 01, 2017 at 07:02 Hours UTC

coveralls commented Nov 30, 2017

coveralls commented Nov 30, 2017

takashioya commented Nov 30, 2017

coveralls commented Nov 30, 2017

coveralls commented Nov 30, 2017

rasbt commented Nov 30, 2017

rasbt left a comment

Choose a reason for hiding this comment

rasbt Nov 30, 2017

Choose a reason for hiding this comment

rasbt Nov 30, 2017

Choose a reason for hiding this comment

rasbt Nov 30, 2017

Choose a reason for hiding this comment

coveralls commented Dec 1, 2017

coveralls commented Dec 1, 2017

coveralls commented Dec 1, 2017

coveralls commented Dec 1, 2017

takashioya commented Nov 30, 2017 •

edited

Loading

pep8speaks commented Nov 30, 2017 •

edited

Loading