DOC highlights: stacking estimators #15414
Conversation
|
@thebooort sorry I saw your message too late :/ |
| from sklearn.ensemble import StackingClassifier | ||
| from sklearn.model_selection import train_test_split | ||
|
|
||
| X, y = load_iris(return_X_y=True) |
There was a problem hiding this comment.
For the docstring, it was enough but do you think that in the highlight we should use another dataset than iris?
There was a problem hiding this comment.
I think the highlights should be as easy as the docstrings, and the user guide is where the examples should be maybe more meaningful. We can link to the userguide for more info though.
| # allows to use the strength of each individual estimator by using their output | ||
| # as input of a final estimator. | ||
| # Note that ``estimators_`` are fitted on the full ``X`` while | ||
| # ``final_estimator``_ is trained using cross-validated predictions of the |
There was a problem hiding this comment.
| # ``final_estimator``_ is trained using cross-validated predictions of the | |
| # ``final_estimator_`` is trained using cross-validated predictions of the |
| ############################################################################ | ||
| # Stacking Classifier (and Regressor) | ||
| # ----------------------------------- | ||
| # These new estimators are a stack of estimators with a final classifier or |
There was a problem hiding this comment.
Remove the first line and link to estimators?
StackingClassifer and StackingRegressor implements stacked generalization, which consist of stacking the output...
| @@ -12,7 +12,7 @@ | |||
|
|
|||
| To install the latest version (with pip):: | |||
|
|
|||
| pip install -U scikit-learn --upgrade | |||
| pip install -U scikit-learn | |||
There was a problem hiding this comment.
I'm confused by this. As I understand, -U is the same as --upgrade, isn't it?
There was a problem hiding this comment.
lol I thought it was for --user (which I've never ever used)
but you're right, let's keep --upgrade and remove -U then
Towards #15152
This is the example we have already in its docstring.