GridSearchQ#

class mlquantify.model_selection.GridSearchQ(quantifier, param_grid, protocol='app', samples_sizes=100, n_repetitions=10, scoring=<function MAE>, refit=True, val_split=0.4, n_jobs=1, random_seed=42, verbose=False)[source]#

Grid Search for Quantifiers with evaluation protocols.

This class automates the hyperparameter search over a grid of parameter combinations for a given quantifier. It evaluates each combination using a specified evaluation protocol (e.g., APP, NPP, UPP), over multiple splits of the validation data, and selects the best-performing parameters based on a chosen scoring metric such as Mean Absolute Error (MAE).

Parameters:

quantifierBaseQuantifier: Quantifier class (not instance). It must implement fit and predict.
param_griddict: Dictionary where keys are parameter names and values are lists of parameter values to try.
protocol{‘app’, ‘npp’, ‘upp’}, default=’app’: Evaluation protocol to use for splitting the validation data.
samples_sizesint or list of int, default=100: Batch size(s) for evaluation splits.
n_repetitionsint, default=10: Number of random repetitions per evaluation.
scoringcallable, default=MAE: Scoring function to evaluate prevalence prediction quality. Must accept (true_prevalences, predicted_prevalences) arrays.
refitbool, default=True: If True, refits the quantifier on the whole data using best parameters.
val_splitfloat, default=0.4: Fraction of data reserved for validation during parameter search.
n_jobsint or None, default=1: Number of parallel jobs for evaluation.
random_seedint, default=42: Random seed for reproducibility.
verbosebool, default=False: Enable verbose output during evaluation.

Attributes:

best_scorefloat: Best score (lowest loss) found during grid search.
best_paramsdict: Return the best parameters found during fitting.
best_model_BaseQuantifier: Refitted quantifier instance with best parameters after search.

Methods

fit(X, y)	Runs grid search over param_grid, evaluates with the selected protocol, and stores best found parameters and model.
predict(X)	Predicts prevalences using the best fitted model after search.
best_params()	Returns the best parameter dictionary after fitting.
best_model()	Returns the best refitted quantifier after fitting.
sout(msg)	Utility method to print messages if verbose is enabled.

Examples

>>> from mlquantify.quantifiers import SomeQuantifier
>>> param_grid = {'alpha': [0.1, 1.0], 'beta': [10, 20]}
>>> grid_search = GridSearchQ(quantifier=SomeQuantifier,
...                          param_grid=param_grid,
...                          protocol='app',
...                          samples_sizes=100,
...                          n_repetitions=5,
...                          scoring=MAE,
...                          refit=True,
...                          val_split=0.3,
...                          n_jobs=2,
...                          random_seed=123,
...                          verbose=True)
>>> grid_search.fit(X_train, y_train)
>>> y_pred = grid_search.predict(X_test)
>>> best_params = grid_search.best_params()
>>> best_model = grid_search.best_model()

best_model()[source]#

Return the best model after fitting.

Returns:

Quantifier: The best fitted model.

Raises:

ValueError: If called before fitting.

best_params()[source]#

Return the best parameters found during fitting.

Returns:

dict: The best parameters.

Raises:

ValueError: If called before fitting.

fit(X, y)[source]#

Fit quantifiers over grid parameter combinations with evaluation protocol.

Splits data into training and validation by val_split, and evaluates each parameter combination multiple times with protocol-generated batches.

Parameters:

Xarray-like: Feature matrix for training.
yarray-like: Target labels for training.

Returns:

selfobject: Returns self for chaining.

get_metadata_routing()[source]#

Get metadata routing of this object.

Please check User Guide on how the routing mechanism works.

Returns:

routingMetadataRequest: A MetadataRequest encapsulating routing information.

get_params(deep=True)[source]#

Get parameters for this estimator.

Parameters:

deepbool, default=True: If True, will return the parameters for this estimator and contained subobjects that are estimators.

Returns:

paramsdict: Parameter names mapped to their values.

predict(X)[source]#

Predict using the best found model.

Parameters:

Xarray-like: Data for prediction.

Returns:

predictionsarray-like: Prevalence predictions.

Raises:

RuntimeError: If called before fitting.

save_quantifier(path: str | None = None) → None[source]#: Save the quantifier instance to a file.

set_params(**params)[source]#

Set the parameters of this estimator.

The method works on simple estimators as well as on nested objects (such as Pipeline). The latter have parameters of the form <component>__<parameter> so that it’s possible to update each component of a nested object.

Parameters:

**paramsdict: Estimator parameters.

Returns:

selfestimator instance: Estimator instance.

sout(msg)[source]#: Prints messages if verbose is True.

GridSearchQ#

This Page