skrub.Expr.skb.get_grid_search

Expr.skb.get_grid_search(*, fitted=False, **kwargs)

Find the best parameters with grid search.

This function returns a ParamSearch, an object similar to scikit-learn’s GridSearchCV. The main difference is that methods such as fit() and predict() accept a dictionary of inputs rather than X and y. Please refer to the examples gallery for an in-depth explanation.

If the expression contains some numeric ranges (choose_float, choose_int), either discretize them by providing the n_steps argument or use get_randomized_search instead of get_grid_search.

Parameters:

fittedbool (default=False)

If True, the gridsearch is fitted on the data provided when initializing variables in this expression (the data returned by .skb.get_data()).

kwargsdict

All other named arguments are forwarded to sklearn.search.GridSearchCV.

Returns:

ParamSearch

An object implementing the hyperparameter search. Besides the usual fit, predict, attributes of interest are results_ and plot_results().

See also

skrub.Expr.skb.get_randomized_search

Find the best parameters with grid search.

Examples

>>> import skrub
>>> from sklearn.datasets import make_classification
>>> from sklearn.linear_model import LogisticRegression
>>> from sklearn.ensemble import RandomForestClassifier
>>> from sklearn.dummy import DummyClassifier

>>> X_a, y_a = make_classification(random_state=0)
>>> X, y = skrub.X(X_a), skrub.y(y_a)
>>> logistic = LogisticRegression(C=skrub.choose_from([0.1, 10.0], name="C"))
>>> rf = RandomForestClassifier(
...     n_estimators=skrub.choose_from([3, 30], name="N 🌴"),
...     random_state=0,
... )
>>> classifier = skrub.choose_from(
...     {"logistic": logistic, "rf": rf, "dummy": DummyClassifier()}, name="classifier"
... )
>>> pred = X.skb.apply(classifier, y=y)
>>> print(pred.skb.describe_param_grid())
- classifier: 'logistic'
  C: [0.1, 10.0]
- classifier: 'rf'
  N 🌴: [3, 30]
- classifier: 'dummy'

>>> search = pred.skb.get_grid_search(fitted=True)
>>> search.results_
   mean_test_score     C   N 🌴 classifier
0             0.89   NaN  30.0         rf
1             0.84   0.1   NaN   logistic
2             0.80  10.0   NaN   logistic
3             0.65   NaN   3.0         rf
4             0.50   NaN   NaN      dummy