def transform(self, X):
"""Transform X separately by each transformer, concatenate results.
Parameters
----------
X : iterable or array-like, depending on transformers
Input data to be transformed.
Returns
-------
X_t : array-like or sparse matrix, shape (n_samples, sum_n_components)
hstack of results of transformers. sum_n_components is the
sum of n_components (output dimension) over transformers.
"""
Xs = Parallel(n_jobs=self.n_jobs)(
delayed(_transform_one)(trans, X, None, weight)
for name, trans, weight in self._iter())
if not Xs:
# All transformers are None
return np.zeros((X.shape[0], 0))
if any(sparse.issparse(f) for f in Xs):
Xs = sparse.vstack(Xs).tocsr()
else:
if isinstance(Xs[0], np.ndarray):
Xs = np.vstack(Xs)
elif isinstance(Xs[0], pd.Series) or isinstance(
Xs[0], pd.DataFrame):
Xs = pd.concat(Xs, axis=1)
return Xs
def _resample_model(estimator_func,
X,
y,
scaling=.5,
n_resampling=200,
n_jobs=None,
verbose=False,
pre_dispatch='3*n_jobs',
random_state=None,
sample_fraction=.75,
**params):
random_state = check_random_state(random_state)
# We are generating 1 - weights, and not weights
n_samples, n_features = X.shape
if not (0 < scaling < 1):
raise ValueError(
"'scaling' should be between 0 and 1. Got %r instead." % scaling)
scaling = 1. - scaling
scores_ = 0.0
for active_set in Parallel(
n_jobs=n_jobs, verbose=verbose,
pre_dispatch=pre_dispatch)(delayed(estimator_func)(
X,
y,
weights=scaling *
random_state.randint(0, 2, size=(n_features, )),
mask=(random_state.rand(n_samples) < sample_fraction),
verbose=max(0, verbose - 1),
**params) for _ in range(n_resampling)):
scores_ += active_set
scores_ /= n_resampling
return scores_
def fit(self, X, y, categories="auto"):
# this is hard-coded for categorical variables
if isinstance(y, pd.Series) and hasattr(y, "cat"):
y = y.cat.codes
self.n_classes_ = np.max(y) + 1
categories = list(range(self.n_classes_))
# order of estimators
self.estimators_ = Parallel(n_jobs=self.n_jobs)(
delayed(_parallel_fit_estimator)(clone(self.estimator), X, y, cat)
for cat in categories[:-1])
return self
def validation_curve(estimator,
X,
y,
param_name,
param_range,
groups=None,
cv=None,
scoring=None,
n_jobs=None,
pre_dispatch="all",
verbose=0,
error_score=np.nan):
X, y, groups = indexable(X, y, groups)
cv = check_cv(cv, y, classifier=is_classifier(estimator))
scorer = check_scoring(estimator, scoring=scoring)
parallel = Parallel(n_jobs=n_jobs,
pre_dispatch=pre_dispatch,
verbose=verbose)
out = parallel(
delayed(_fit_and_score)(clone(estimator),
X,
y,
scorer,
train,
test,
verbose,
parameters={
param_name: v
},
fit_params=None,
return_train_score=True,
error_score=error_score,
return_estimator=True,
return_times=True)
# NOTE do not change order of iteration to allow one time cv splitters
for train, test in cv.split(X, y, groups) for v in param_range)
out = np.asarray(out)
estimators = out[:, 4]
out_scores = np.asarray(out[:, :2])
fit_time = out[:, 2]
score_time = out[:, 3]
n_params = len(param_range)
n_cv_folds = out_scores.shape[0] // n_params
out_scores = out_scores.reshape(n_cv_folds, n_params, 2).transpose(
(2, 1, 0))
return estimators, np.float64(out_scores[0]), np.float64(out_scores[1]), np.float64(fit_time), \
np.float64(score_time)
def fit(self, X, y):
# n_samples, n_features = X.shape
X = self._augment(X)
# Perform label encoding so label indicies start from zero
le = LabelEncoder()
encoded_y = le.fit_transform(y)
self.classes_ = le.classes_
n_classes = len(self.classes_)
# Use the Parallel library to fit C binary classifiers in parallel
results = Parallel(
n_jobs=self.n_jobs, prefer='threads',
verbose=self.verbose)(delayed(_fit_binary_perceptron)(
X, encoded_y, c, self.eta0, self.decay, self.max_iterations)
for c in range(n_classes))
# Store final result for prediction
self.weights_ = np.array(results)
return self
def transform (self, X): """Transform X separately by each transformer, concatenate results. Parameters ---------- X : iterable or array-like, depending on transformers Input data to be transformed. Returns ------- X_t : array-like or sparse matrix, shape (n_samples, sum_n_components) hstack of results of transformers. sum_n_components is the sum of n_components (output dimension) over transformers. """ if self.use_in_model: return super(CustomFeatureUnion, self).transform(X) Xs = Parallel(n_jobs=self.n_jobs)( delayed(_transform_one)(trans, X, None, weight) for _, trans, weight in self._iter()) return self.get_result_as_dictionary(Xs)
def fit(self, X, y=None):
"""Fit all transformers using X.
Parameters
----------
X : iterable or array-like, depending on transformers
Input data, used to fit transformers.
y : array-like, shape (n_samples, ...), optional
Targets for supervised learning.
Returns
-------
self : FeatureUnion
This estimator
"""
self.transformer_list = list(self.transformer_list)
self._validate_transformers()
transformers = Parallel(n_jobs=self.n_jobs)(
delayed(_fit_one_transformer)(trans, X, y)
for _, trans, _ in self._iter())
self._update_transformer_list(transformers)
return self
def fit_transform(self, X, y=None, **fit_params):
"""Fit all transformers, transform the data and concatenate results.
Parameters
----------
X : iterable or array-like, depending on transformers
Input data to be transformed.
y : array-like, shape (n_samples, ...), optional
Targets for supervised learning.
Returns
-------
X_t : array-like or sparse matrix, shape (n_samples, sum_n_components)
hstack of results of transformers. sum_n_components is the
sum of n_components (output dimension) over transformers.
"""
self._validate_transformers()
result = Parallel(n_jobs=self.n_jobs)(
delayed(_fit_transform_one)(trans, X, y, weight, **fit_params)
for name, trans, weight in self._iter())
if not result:
# All transformers are None
return np.zeros((X.shape[0], 0))
Xs, transformers = zip(*result)
self._update_transformer_list(transformers)
if any(sparse.issparse(f) for f in Xs):
Xs = sparse.vstack(Xs).tocsr()
else:
if isinstance(Xs[0], np.ndarray):
Xs = np.vstack(Xs)
elif isinstance(Xs[0], pd.Series) or isinstance(
Xs[0], pd.DataFrame):
Xs = pd.concat(Xs, axis=1)
return Xs
def fit(self, X, y, sample_weight=None):
""" Fit the estimators.
Parameters
----------
X : {array-like, sparse matrix}, shape = [n_samples, n_features]
Training vectors, where n_samples is the number of samples and
n_features is the number of features.
y : array-like, shape = [n_samples]
Target values.
sample_weight : array-like, shape = [n_samples] or None
Sample weights. If None, then samples are equally weighted.
Note that this is supported only if all underlying estimators
support sample weights.
Returns
-------
self : object
"""
if isinstance(y, np.ndarray) and len(y.shape) > 1 and y.shape[1] > 1:
raise NotImplementedError('Multilabel and multi-output'
' classification is not supported.')
if self.voting not in ('soft', 'hard'):
raise ValueError("Voting must be 'soft' or 'hard'; got (voting=%r)"
% self.voting)
if self.estimators is None or len(self.estimators) == 0:
raise AttributeError('Invalid `estimators` attribute, `estimators`'
' should be a list of (string, estimator)'
' tuples')
if (self.weights is not None and
len(self.weights) != len(self.estimators)):
raise ValueError('Number of classifiers and weights must be equal'
'; got %d weights, %d estimators'
% (len(self.weights), len(self.estimators)))
if sample_weight is not None:
for name, step in self.estimators:
if not has_fit_parameter(step, 'sample_weight'):
raise ValueError('Underlying estimator \'%s\' does not'
' support sample weights.' % name)
names, clfs = zip(*self.estimators)
self._validate_names(names)
n_isnone = np.sum([clf is None for _, clf in self.estimators])
if n_isnone == len(self.estimators):
raise ValueError('All estimators are None. At least one is '
'required to be a classifier!')
self.le_ = LabelEncoder().fit(y)
self.classes_ = self.le_.classes_
self.estimators_ = []
transformed_y = self.le_.transform(y)
self.estimators_ = Parallel(n_jobs=self.n_jobs)(
delayed(_parallel_fit_estimator)(clone(clf), X, transformed_y,
sample_weight=sample_weight)
for clf in clfs if clf is not None)
self.named_estimators_ = Bunch(**dict())
for k, e in zip(self.estimators, self.estimators_):
self.named_estimators_[k[0]] = e
return self
def fit(self, X, y=None):
"""Fits the GraphLasso covariance model to X.
Parameters
----------
X : ndarray, shape (n_samples, n_features)
Data from which to compute the covariance estimate
y : (ignored)
"""
# Covariance does not make sense for a single feature
self.random_state = check_random_state(self.random_state)
# check_data_dimensions(X, layers=2)
X = check_array(X, ensure_min_features=2,
ensure_min_samples=2, estimator=self)
self.X_train = X
if self.assume_centered:
self.location_ = np.zeros((X.shape[0], X.shape[1]))
else:
self.location_ = X.mean(0)
emp_cov = empirical_covariance(
X, assume_centered=self.assume_centered)
X = check_array(X, ensure_min_features=2, estimator=self)
cv = check_cv(self.cv, y, classifier=False)
# List of (alpha, scores, covs)
path = list()
n_etas = self.etas
inner_verbose = max(0, self.verbose - 1)
if isinstance(n_etas, Sequence):
etas = self.etas
else:
eta_1 = par_max(emp_cov)
eta_0 = 1e-2 * eta_1
etas = np.logspace(np.log10(eta_0), np.log10(eta_1),
n_etas)[::-1]
n_mus = self.mus
inner_verbose = max(0, self.verbose - 1)
if isinstance(n_mus, Sequence):
mus = self.mus
else:
mu_1 = par_max(emp_cov) # not sure is the best strategy
mu_0 = 1e-2 * mu_1
mus = np.logspace(np.log10(mu_0), np.log10(mu_1),
n_mus)[::-1]
with warnings.catch_warnings():
warnings.simplefilter('ignore', ConvergenceWarning)
this_path = Parallel(
n_jobs=self.n_jobs,
verbose=self.verbose
)(delayed(flgl_path)(X[train], links=self.links, etas=etas, mus= mus,
X_test=X[test], tol=self.tol,
max_iter=int(.1 * self.max_iter),
update_rho=self.update_rho,
verbose=0, random_state=self.random_state)
for train, test in cv.split(X, y))
# Little danse to transform the list in what we need
covs, precs, hidds, scores = zip(*this_path)
covs = zip(*covs)
precs = zip(*precs)
hidds = zip(*hidds)
scores = zip(*scores)
combinations = list(product(etas, mus))
path.extend(zip(combinations, scores, covs))
path = sorted(path, key=operator.itemgetter(0), reverse=True)
# Find the maximum (avoid using built in 'max' function to
# have a fully-reproducible selection of the smallest alpha
# in case of equality)
best_score = -np.inf
last_finite_idx = 0
for index, (combination, scores, _) in enumerate(path):
this_score = np.mean(scores)
if this_score >= .1 / np.finfo(np.float64).eps:
this_score = np.nan
if np.isfinite(this_score):
last_finite_idx = index
if this_score >= best_score:
best_score = this_score
best_index = index
path = list(zip(*path))
grid_scores = list(path[1])
parameters = list(path[0])
# Finally, compute the score with alpha = 0
best_eta, best_mu = combinations[best_index]
self.eta_ = best_eta
self.mu_ = best_mu
self.cv_parameters_ = combinations
# Finally fit the model with the selected alpha
self.covariance_, self.precision_, self.hidden_, self.R_, self.n_iter_ = two_layers_fixed_links_GL(
emp_cov, self.links, eta=best_eta, mu=best_mu, tol=self.tol,
max_iter=self.max_iter,
verbose=self.verbose, random_state=self.random_state,
compute_objective=True, return_n_iter=True)
return self
def __call__(self, value):
values = Parallel(n_jobs=self.n_jobs)(delayed(pfn)(value)
for pfn in self.steps)
return self.aggregate(values)
def kneighbors(self, X, n_neighbors=None, return_distance=True):
"""Finds the K-neighbors of a point.
Returns indices of and distances to the neighbors of each point.
Parameters
----------
X : sktime-format pandas dataframe with shape([n_cases,n_dimensions]),
or numpy ndarray with shape([n_cases,n_readings,n_dimensions])
y : {array-like, sparse matrix}
Target values of shape = [n_samples]
n_neighbors : int
Number of neighbors to get (default is the value
passed to the constructor).
return_distance : boolean, optional. Defaults to True.
If False, distances will not be returned
Returns
-------
dist : array
Array representing the lengths to points, only present if
return_distance=True
ind : array
Indices of the nearest points in the population matrix.
Examples
--------
In the following example, we construct a NeighborsClassifier
class from an array representing our data set and ask who's
the closest point to [1,1,1]
>>> samples = [[0., 0., 0.], [0., .5, 0.], [1., 1., .5]]
>>> from sklearn.neighbors import NearestNeighbors
>>> neigh = NearestNeighbors(n_neighbors=1)
>>> neigh.fit(samples) # doctest: +ELLIPSIS
NearestNeighbors(algorithm='auto', leaf_size=30, ...)
>>> print(neigh.kneighbors([[1., 1., 1.]])) # doctest: +ELLIPSIS
(array([[0.5]]), array([[2]]))
As you can see, it returns [[0.5]], and [[2]], which means that the
element is at distance 0.5 and is the third element of samples
(indexes start at 0). You can also query for multiple points:
>>> X = [[0., 1., 0.], [1., 0., 1.]]
>>> neigh.kneighbors(X, return_distance=False) # doctest: +ELLIPSIS
array([[1],
[2]]...)
"""
check_data_sktime_tsc(X)
check_is_fitted(self, "_fit_method")
if n_neighbors is None:
n_neighbors = self.n_neighbors
elif n_neighbors <= 0:
raise ValueError("Expected n_neighbors > 0. Got %d" % n_neighbors)
else:
if not np.issubdtype(type(n_neighbors), np.integer):
raise TypeError("n_neighbors does not take %s value, "
"enter integer value" % type(n_neighbors))
if X is not None:
query_is_train = False
X = check_array(X, accept_sparse='csr', allow_nd=True)
else:
query_is_train = True
X = self._fit_X
# Include an extra neighbor to account for the sample itself being
# returned, which is removed later
n_neighbors += 1
train_size = self._fit_X.shape[0]
if n_neighbors > train_size:
raise ValueError("Expected n_neighbors <= n_samples, "
" but n_samples = %d, n_neighbors = %d" %
(train_size, n_neighbors))
n_samples = X.shape[0]
sample_range = np.arange(n_samples)[:, None]
n_jobs = effective_n_jobs(self.n_jobs)
if self._fit_method == 'brute':
reduce_func = partial(self._kneighbors_reduce_func,
n_neighbors=n_neighbors,
return_distance=return_distance)
# for efficiency, use squared euclidean distances
kwds = ({
'squared': True
} if self.effective_metric_ == 'euclidean' else
self.effective_metric_params_)
result = pairwise_distances_chunked(X,
self._fit_X,
reduce_func=reduce_func,
metric=self.effective_metric_,
n_jobs=n_jobs,
**kwds)
elif self._fit_method in ['ball_tree', 'kd_tree']:
if issparse(X):
raise ValueError(
"%s does not work with sparse matrices. Densify the data, "
"or set algorithm='brute'" % self._fit_method)
if LooseVersion(joblib_version) < LooseVersion('0.12'):
# Deal with change of API in joblib
delayed_query = delayed(self._tree.query, check_pickle=False)
parallel_kwargs = {"backend": "threading"}
else:
delayed_query = delayed(self._tree.query)
parallel_kwargs = {"prefer": "threads"}
result = Parallel(n_jobs, **parallel_kwargs)(
delayed_query(X[s], n_neighbors, return_distance)
for s in gen_even_slices(X.shape[0], n_jobs))
else:
raise ValueError("internal: _fit_method not recognized")
if return_distance:
dist, neigh_ind = zip(*result)
result = np.vstack(dist), np.vstack(neigh_ind)
else:
result = np.vstack(result)
if not query_is_train:
return result
else:
# If the query data is the same as the indexed data, we would like
# to ignore the first nearest neighbor of every sample, i.e
# the sample itself.
if return_distance:
dist, neigh_ind = result
else:
neigh_ind = result
sample_mask = neigh_ind != sample_range
# Corner case: When the number of duplicates are more
# than the number of neighbors, the first NN will not
# be the sample, but a duplicate.
# In that case mask the first duplicate.
dup_gr_nbrs = np.all(sample_mask, axis=1)
sample_mask[:, 0][dup_gr_nbrs] = False
neigh_ind = np.reshape(neigh_ind[sample_mask],
(n_samples, n_neighbors - 1))
if return_distance:
dist = np.reshape(dist[sample_mask],
(n_samples, n_neighbors - 1))
return dist, neigh_ind
return neigh_ind
def fit(self, X, y=None, groups=None, **fit_params):
"""Run fit with all sets of parameters.
Parameters
----------
X : array-like, shape = [n_samples, n_features]
Training vector, where n_samples is the number of samples and
n_features is the number of features.
y : array-like, shape = [n_samples] or [n_samples, n_output], optional
Target relative to X for classification or regression;
None for unsupervised learning.
groups : array-like, with shape (n_samples,), optional
Group labels for the samples used while splitting the dataset into
train/test set.
**fit_params : dict of string -> object
Parameters passed to the ``fit`` method of the estimator
"""
if self.fit_params is not None:
warnings.warn('"fit_params" as a constructor argument was '
'deprecated in version 0.19 and will be removed '
'in version 0.21. Pass fit parameters to the '
'"fit" method instead.', DeprecationWarning)
if fit_params:
warnings.warn('Ignoring fit_params passed as a constructor '
'argument in favor of keyword arguments to '
'the "fit" method.', RuntimeWarning)
else:
fit_params = self.fit_params
estimator = self.estimator
cv = check_cv(self.cv, y, classifier=is_classifier(estimator))
scorers, self.multimetric_ = _check_multimetric_scoring(
self.estimator, scoring=self.scoring)
if self.multimetric_:
if self.refit is not False and (
not isinstance(self.refit, six.string_types) or
# This will work for both dict / list (tuple)
self.refit not in scorers):
raise ValueError("For multi-metric scoring, the parameter "
"refit must be set to a scorer key "
"to refit an estimator with the best "
"parameter setting on the whole data and "
"make the best_* attributes "
"available for that metric. If this is not "
"needed, refit should be set to False "
"explicitly. %r was passed." % self.refit)
else:
refit_metric = self.refit
else:
refit_metric = 'score'
X, y, groups = indexable(X, y, groups)
n_splits = cv.get_n_splits(X, y, groups)
base_estimator = clone(self.estimator)
parallel = Parallel(n_jobs=self.n_jobs, verbose=self.verbose,
pre_dispatch=self.pre_dispatch)
fit_and_score_kwargs = dict(scorer=scorers,
fit_params=fit_params,
return_train_score=self.return_train_score,
return_n_test_samples=True,
return_times=True,
return_parameters=False,
return_estimator=self.return_estimator,
error_score=self.error_score,
verbose=self.verbose)
results_container = [{}]
with parallel:
all_candidate_params = []
all_out = []
all_estimators = []
def evaluate_candidates(candidate_params):
candidate_params = list(candidate_params)
n_candidates = len(candidate_params)
def _fit_and_score_recv(i_fold, X, y, train, test, parameters):
current_estimator = clone(base_estimator)
if isinstance(current_estimator, Pipeline):
if hasattr(current_estimator._final_estimator, 'cachedir'):
current_estimator._final_estimator.cachedir = os.path.join(self.cachedir,
'%i_fold ' % i_fold)
else:
warnings.warn('Final estimator does not have recovery'
' or saving capabilities')
elif hasattr(current_estimator, 'cachedir'):
current_estimator.cachedir = os.path.join(self.cachedir, '%i_fold ' % i_fold)
else:
warnings.warn('Estimator does not have recovery '
' or saving capabilities')
print parameters
print i_fold
return delayed(_fit_and_score)(current_estimator,
X, y,
train=train, test=test,
parameters=parameters,
**fit_and_score_kwargs)
list_split = list(enumerate(cv.split(X, y, groups)))
if self.verbose > 0:
print("Fitting {0} folds for each of {1} candidates,"
" totalling {2} fits".format(
n_splits, n_candidates, n_candidates * n_splits))
# print list(candidate_params)
# raise NotImplementedError
if self.client is None:
out = parallel(_fit_and_score_recv(i_fold,
X, y,
train, test,
parameters)
for (parameters, (i_fold, (train, test)))
in product(candidate_params,
list_split))
else:
self.client[:].use_dill()
dview = self.client[:]
out = dview.map(lambda parameters, i_fold, train, test: _fit_and_score_recv(i_fold,
X, y,
train, test,
parameters),
[(parameters, i_fold, train, test) for (parameters, (i_fold, (train, test)))
in product(candidate_params,
list_split)])
if self.return_estimator:
all_estimators.extend([out_set[-1] for out_set in out])
out = [out_set[:-1] for out_set in out]
all_candidate_params.extend(candidate_params)
all_out.extend(out)
# XXX: When we drop Python 2 support, we can use nonlocal
# instead of results_container
results_container[0] = self._format_results(
all_candidate_params, scorers, n_splits, all_out)
return results_container[0]
self._run_search(evaluate_candidates)
results = results_container[0]
# For multi-metric evaluation, store the best_index_, best_params_ and
# best_score_ iff refit is one of the scorer names
# In single metric evaluation, refit_metric is "score"
if self.refit or not self.multimetric_:
self.best_index_ = results["rank_test_%s" % refit_metric].argmin()
self.best_params_ = results["params"][self.best_index_]
self.best_score_ = results["mean_test_%s" % refit_metric][
self.best_index_]
if self.refit:
self.best_estimator_ = clone(base_estimator).set_params(
**self.best_params_)
refit_start_time = time.time()
if y is not None:
self.best_estimator_.fit(X, y, **fit_params)
else:
self.best_estimator_.fit(X, **fit_params)
refit_end_time = time.time()
self.refit_time_ = refit_end_time - refit_start_time
# Store the only scorer not as a dict for single metric evaluation
self.scorer_ = scorers if self.multimetric_ else scorers['score']
self.cv_results_ = results
self.n_splits_ = n_splits
if self.return_estimator:
self.cv_estimators = all_estimators
return self
def exp(solvers, penalties, single_target, n_samples=30000, max_iter=20,
dataset='rcv1', n_jobs=1, skip_slow=False):
mem = Memory(cachedir=expanduser('~/cache'), verbose=0)
if dataset == 'rcv1':
rcv1 = fetch_rcv1()
lbin = LabelBinarizer()
lbin.fit(rcv1.target_names)
X = rcv1.data
y = rcv1.target
y = lbin.inverse_transform(y)
le = LabelEncoder()
y = le.fit_transform(y)
if single_target:
y_n = y.copy()
y_n[y > 16] = 1
y_n[y <= 16] = 0
y = y_n
elif dataset == 'digits':
digits = load_digits()
X, y = digits.data, digits.target
if single_target:
y_n = y.copy()
y_n[y < 5] = 1
y_n[y >= 5] = 0
y = y_n
elif dataset == 'iris':
iris = load_iris()
X, y = iris.data, iris.target
elif dataset == '20newspaper':
ng = fetch_20newsgroups_vectorized()
X = ng.data
y = ng.target
if single_target:
y_n = y.copy()
y_n[y > 4] = 1
y_n[y <= 16] = 0
y = y_n
X = X[:n_samples]
y = y[:n_samples]
cached_fit = mem.cache(fit_single)
out = Parallel(n_jobs=n_jobs, mmap_mode=None)(
delayed(cached_fit)(solver, X, y,
penalty=penalty, single_target=single_target,
C=1, max_iter=max_iter, skip_slow=skip_slow)
for solver in solvers
for penalty in penalties)
res = []
idx = 0
for solver in solvers:
for penalty in penalties:
if not (skip_slow and solver == 'lightning' and penalty == 'l1'):
lr, times, train_scores, test_scores, accuracies = out[idx]
this_res = dict(solver=solver, penalty=penalty,
single_target=single_target,
times=times, train_scores=train_scores,
test_scores=test_scores,
accuracies=accuracies)
res.append(this_res)
idx += 1
with open('bench_saga.json', 'w+') as f:
json.dump(res, f)
def lasso_stability_path(X,
y,
scaling=0.5,
random_state=None,
n_resampling=200,
n_grid=100,
sample_fraction=0.75,
eps=4 * np.finfo(np.float).eps,
n_jobs=None,
verbose=False):
"""Stability path based on randomized Lasso estimates
Parameters
----------
X : array-like, shape = [n_samples, n_features]
training data.
y : array-like, shape = [n_samples]
target values.
scaling : float, optional, default=0.5
The alpha parameter in the stability selection article used to
randomly scale the features. Should be between 0 and 1.
random_state : int, RandomState instance or None, optional, default=None
The generator used to randomize the design. If int, random_state is
the seed used by the random number generator; If RandomState instance,
random_state is the random number generator; If None, the random number
generator is the RandomState instance used by `np.random`.
n_resampling : int, optional, default=200
Number of randomized models.
n_grid : int, optional, default=100
Number of grid points. The path is linearly reinterpolated
on a grid between 0 and 1 before computing the scores.
sample_fraction : float, optional, default=0.75
The fraction of samples to be used in each randomized design.
Should be between 0 and 1. If 1, all samples are used.
eps : float, optional
Smallest value of alpha / alpha_max considered
n_jobs : int or None, optional (default=None)
Number of CPUs to use during the resampling.
``None`` means 1 unless in a :obj:`joblib.parallel_backend` context.
``-1`` means using all processors. See :term:`Glossary <n_jobs>`
for more details.
verbose : boolean or integer, optional
Sets the verbosity amount
Returns
-------
alphas_grid : array, shape ~ [n_grid]
The grid points between 0 and 1: alpha/alpha_max
scores_path : array, shape = [n_features, n_grid]
The scores for each feature along the path.
"""
X, y = check_X_y(X, y, accept_sparse=['csr', 'csc', 'coo'])
rng = check_random_state(random_state)
if not (0 < scaling < 1):
raise ValueError("Parameter 'scaling' should be between 0 and 1."
" Got %r instead." % scaling)
n_samples, n_features = X.shape
paths = Parallel(n_jobs=n_jobs, verbose=verbose)(
delayed(_lasso_stability_path)(
X,
y,
mask=rng.rand(n_samples) < sample_fraction,
weights=1. - scaling * rng.randint(0, 2, size=(n_features, )),
eps=eps) for k in range(n_resampling))
all_alphas = sorted(list(set(itertools.chain(*[p[0] for p in paths]))))
# Take approximately n_grid values
stride = int(max(1, int(len(all_alphas) / float(n_grid))))
all_alphas = all_alphas[::stride]
if not all_alphas[-1] == 1:
all_alphas.append(1.)
all_alphas = np.array(all_alphas)
scores_path = np.zeros((n_features, len(all_alphas)))
for alphas, coefs in paths:
if alphas[0] != 0:
alphas = np.r_[0, alphas]
coefs = np.c_[np.ones((n_features, 1)), coefs]
if alphas[-1] != all_alphas[-1]:
alphas = np.r_[alphas, all_alphas[-1]]
coefs = np.c_[coefs, np.zeros((n_features, 1))]
scores_path += (interp1d(alphas,
coefs,
kind='nearest',
bounds_error=False,
fill_value=0,
axis=-1)(all_alphas) != 0)
scores_path /= n_resampling
return all_alphas, scores_path