SearchCV

Inspector functionality specific to estimators that inherit from BaseSearchCV

import sklearn.datasets
from model_inspector import get_inspector
from sklearn.model_selection import GridSearchCV, RandomizedSearchCV, KFold
from sklearn.tree import DecisionTreeRegressor

X, y = sklearn.datasets.load_breast_cancer(return_X_y=True, as_frame=True)

grid = GridSearchCV(
    estimator=DecisionTreeRegressor(),
    param_grid={
        "min_samples_leaf": (1, 2, 5, 10, 20, 50),
        "max_features": (0.3, 0.5, 1.0),
        "splitter": ("best", "random"),
    },
    n_jobs=-1,
    return_train_score=True,
    cv=KFold(5, shuffle=True),
).fit(X, y)

inspector = get_inspector(grid, X, y)

---

source

_SearchCVInspector.plot_scores_vs_hparam

 _SearchCVInspector.plot_scores_vs_hparam (hparam:Optional[str]=None,
                                           score_cols:Union[str,Sequence[s
                                           tr],NoneType]=None, ax=None)

Plot model scores against values of one hyperparameter

Parameters:

• hparam: Name of the hyperparameter to plot against. Must be provided if there are multiple hyperparameters. Any other hyperparameters will be fixed at the value they have in self.model.best_params_.
• score_cols: Name of score columns to plot. By default will be the mean test and (if present) train score for the primary scoring metric.
• ax: Matplotlib Axes object. Plot will be added to this object if provided; otherwise a new Axes object will be generated.

ax = inspector.plot_scores_vs_hparam("min_samples_leaf")

ax = inspector.plot_scores_vs_hparam("max_features")

ax = inspector.plot_scores_vs_hparam("splitter")

---

source

_SearchCVInspector.show_score_vs_hparam_pair

 _SearchCVInspector.show_score_vs_hparam_pair (hparams=None,
                                               score_col=None,
                                               cmap:str|Colormap='PuBu',
                                               low:float=0, high:float=0,
                                               axis:Axis|None=0,
                                               subset:Subset|None=None, te
                                               xt_color_threshold:float=0.
                                               408, vmin:float|None=None,
                                               vmax:float|None=None,
                                               gmap:Sequence|None=None)

Show model scores against a pair of hyperparameters.

Background gradient uses axis=None by default, to facilitate identifying the best score across all combinations of hyperparameter values.

Parameters:

• hparams: Name of the hyperparameters to plot against. The first two hyperparameters in self.model.param_grid will be used by default. Any other hyperparameters will be fixed at the value they have in self.model.best_params_.
• score_col: Name of score column to plot. By default will be the mean test score for the primary scoring metric.

Remaining parameters are passed to pandas.io.formats.style.background_gradient.

inspector.show_score_vs_hparam_pair(["min_samples_leaf", "max_features"])

param_max_features	0.300000	0.500000	1.000000
param_min_samples_leaf
1	0.702171	0.679882	0.696289
2	0.736160	0.722219	0.742642
5	0.788082	0.766575	0.779670
10	0.754099	0.768806	0.771306
20	0.723585	0.728852	0.746209
50	0.614502	0.671836	0.686244