ml_grid.pipeline.hyperparameter_search

Classes

`PatchedBayesSearchCV`	Bayesian optimization over hyper parameters.
`HyperparameterSearch`	Initializes the HyperparameterSearch class.

Module Contents

class ml_grid.pipeline.hyperparameter_search.PatchedBayesSearchCV(estimator, search_spaces, optimizer_kwargs=None, n_iter=50, scoring=None, fit_params=None, n_jobs=1, n_points=1, iid='deprecated', refit=True, cv=None, verbose=0, pre_dispatch='2*n_jobs', random_state=None, error_score='raise', return_train_score=False)[source]

Bases: skopt.BayesSearchCV

Bayesian optimization over hyper parameters.

BayesSearchCV implements a “fit” and a “score” method. It also implements “predict”, “predict_proba”, “decision_function”, “transform” and “inverse_transform” if they are implemented in the estimator used.

The parameters of the estimator used to apply these methods are optimized by cross-validated search over parameter settings.

In contrast to GridSearchCV, not all parameter values are tried out, but rather a fixed number of parameter settings is sampled from the specified distributions. The number of parameter settings that are tried is given by n_iter.

Parameters are presented as a list of skopt.space.Dimension objects.

Parameters:

estimator (estimator object.) – A object of that type is instantiated for each search point. This object is assumed to implement the scikit-learn estimator api. Either estimator needs to provide a score function, or scoring must be passed.
search_spaces (dict, list of dict or list of tuple containing (dict, int).) – One of these cases: 1. dictionary, where keys are parameter names (strings) and values are skopt.space.Dimension instances (Real, Integer or Categorical) or any other valid value that defines skopt dimension (see skopt.Optimizer docs). Represents search space over parameters of the provided estimator. 2. list of dictionaries: a list of dictionaries, where every dictionary fits the description given in case 1 above. If a list of dictionary objects is given, then the search is performed sequentially for every parameter space with maximum number of evaluations set to self.n_iter. 3. list of (dict, int > 0): an extension of case 2 above, where first element of every tuple is a dictionary representing some search subspace, similarly as in case 2, and second element is a number of iterations that will be spent optimizing over this subspace.
n_iter (int, default=50) – Number of parameter settings that are sampled. n_iter trades off runtime vs quality of the solution. Consider increasing n_points if you want to try more parameter settings in parallel.
optimizer_kwargs (dict, optional) – Dict of arguments passed to Optimizer. For example, {'base_estimator': 'RF'} would use a Random Forest surrogate instead of the default Gaussian Process.
scoring (str, callable, list, tuple or dict, default=None) –
Strategy to evaluate the performance of the cross-validated model on the test set. If None, the score method of the estimator is used. If scoring represents a single score, one can use:
- a single string (see The scoring parameter: defining model evaluation rules);
- a callable (see scoring) that returns a single value.
If scoring represents multiple scores, one can use:
- a list or tuple of unique strings;
- a callable returning a dictionary where the keys are the metric names and the values are the metric scores;
- a dictionary with metric names as keys and callables a values.
Callables must have the signature scorer(estimator, X, y=None)
fit_params (dict, optional) – Parameters to pass to the fit method.
n_jobs (int, default=1) – Number of jobs to run in parallel. At maximum there are n_points times cv jobs available during each iteration.
n_points (int, default=1) – Number of parameter settings to sample in parallel. If this does not align with n_iter, the last iteration will sample less points. See also ask()
pre_dispatch (int, or string, optional) –
Controls the number of jobs that get dispatched during parallel execution. Reducing this number can be useful to avoid an explosion of memory consumption when more jobs get dispatched than CPUs can process. This parameter can be:
- None, in which case all the jobs are immediately created and spawned. Use this for lightweight and fast-running jobs, to avoid delays due to on-demand spawning of the jobs
- An int, giving the exact number of total jobs that are spawned
- A string, giving an expression as a function of n_jobs, as in ‘2*n_jobs’
cv (int, cross-validation generator or an iterable, optional) –
Determines the cross-validation splitting strategy. Possible inputs for cv are:
- None, to use the default 3-fold cross validation,
- integer, to specify the number of folds in a (Stratified)KFold,
- An object to be used as a cross-validation generator.
- An iterable yielding train, test splits.
For integer/None inputs, if the estimator is a classifier and y is either binary or multiclass, StratifiedKFold is used. In all other cases, KFold is used.
refit (bool, str, default=True) – Refit the best estimator with the entire dataset. If “False”, it is impossible to make predictions using this BayesSearchCV instance after fitting. For multiple metric evaluation, this needs to be a str denoting the scorer that would be used to direct the optimization process, and find the best parameters for refitting the estimator at the end.
verbose (integer) – Controls the verbosity: the higher, the more messages.
random_state (int or RandomState) – Pseudo random number generator state used for random uniform sampling from lists of possible values instead of scipy.stats distributions.
error_score ('raise' (default) or numeric) – Value to assign to the score if an error occurs in estimator fitting. If set to ‘raise’, the error is raised. If a numeric value is given, FitFailedWarning is raised. This parameter does not affect the refit step, which will always raise the error.
return_train_score (boolean, default=False) – If 'True', the cv_results_ attribute will include training scores.

Examples

>>> from skopt import BayesSearchCV
>>> # parameter ranges are specified by one of below
>>> from skopt.space import Real, Categorical, Integer
>>>
>>> from sklearn.datasets import load_iris
>>> from sklearn.svm import SVC
>>> from sklearn.model_selection import train_test_split
>>>
>>> X, y = load_iris(return_X_y=True)
>>> X_train, X_test, y_train, y_test = train_test_split(X, y,
...                                                     train_size=0.75,
...                                                     random_state=0)
>>>
>>> # log-uniform: understand as search over p = exp(x) by varying x
>>> opt = BayesSearchCV(
...     SVC(),
...     {
...         'C': Real(1e-6, 1e+6, prior='log-uniform'),
...         'gamma': Real(1e-6, 1e+1, prior='log-uniform'),
...         'degree': Integer(1,8),
...         'kernel': Categorical(['linear', 'poly', 'rbf']),
...     },
...     n_iter=10, n_jobs=-1,
...     random_state=0
... )
>>>
>>> # executes bayesian optimization
>>> _ = opt.fit(X_train, y_train)
>>>
>>> # model can be saved, used for predictions or scoring
>>> print(opt.score(X_test, y_test))
0.973...

cv_results_

A dict with keys as column headers and values as columns, that can be imported into a pandas DataFrame.

For instance the below given table

param_kernel	param_gamma	split0_test_score	…	rank_test_score
‘rbf’	0.1	0.8	…	2
‘rbf’	0.2	0.9	…	1
‘rbf’	0.3	0.7	…	1

will be represented by a cv_results_ dict of:

{
'param_kernel' : masked_array(data = ['rbf', 'rbf', 'rbf'],
                              mask = False),
'param_gamma'  : masked_array(data = [0.1 0.2 0.3], mask = False),
'split0_test_score'  : [0.8, 0.9, 0.7],
'split1_test_score'  : [0.82, 0.5, 0.7],
'mean_test_score'    : [0.81, 0.7, 0.7],
'std_test_score'     : [0.02, 0.2, 0.],
'rank_test_score'    : [3, 1, 1],
'split0_train_score' : [0.8, 0.9, 0.7],
'split1_train_score' : [0.82, 0.5, 0.7],
'mean_train_score'   : [0.81, 0.7, 0.7],
'std_train_score'    : [0.03, 0.03, 0.04],
'mean_fit_time'      : [0.73, 0.63, 0.43, 0.49],
'std_fit_time'       : [0.01, 0.02, 0.01, 0.01],
'mean_score_time'    : [0.007, 0.06, 0.04, 0.04],
'std_score_time'     : [0.001, 0.002, 0.003, 0.005],
'params' : [{'kernel' : 'rbf', 'gamma' : 0.1}, ...],
}

NOTE that the key 'params' is used to store a list of parameter settings dict for all the parameter candidates.

The mean_fit_time, std_fit_time, mean_score_time and std_score_time are all in seconds.

Type:: dict of numpy (masked) ndarrays

best_estimator_

Estimator that was chosen by the search, i.e. estimator which gave highest score (or smallest loss if specified) on the left out data. Not available if refit=False.

Type:: estimator

optimizer_results_[source]

Contains a OptimizeResult for each search space. The search space parameter are sorted by its name.

Type:: list of OptimizeResult

best_score_

Score of best_estimator on the left out data.

Type:: float

best_params_

Parameter setting that gave the best results on the hold out data.

Type:: dict

best_index_

The index (of the cv_results_ arrays) which corresponds to the best candidate parameter setting.

The dict at search.cv_results_['params'][search.best_index_] gives the parameter setting for the best model, that gives the highest mean score (search.best_score_).

Type:: int

scorer_

Scorer function used on the held out data to choose the best parameters for the model.

Type:: function

n_splits_

The number of cross-validation splits (folds/iterations).

Type:: int

refit_time_

Seconds used for refitting the best model on the whole dataset. This is present only if refit is not False.

Type:: float

multimetric_

Whether or not the scorers compute several metrics.

Type:: bool

Notes

The parameters selected are those that maximize the score of the held-out data, according to the scoring parameter.

If n_jobs was set to a value higher than one, the data is copied for each parameter setting (and not n_jobs times). This is done for efficiency reasons if individual jobs take very little time, but may raise errors if the dataset is large and not enough memory is available. A workaround in this case is to set pre_dispatch. Then, the memory is copied only pre_dispatch many times. A reasonable value for pre_dispatch is 2 * n_jobs.