def plot_all_roc_curve(self, splitter=None, num_splits=None):
"plot ROC curves for all estimators"
(clf_list, clf_names) = self.get_component_classifiers()
ensemble_classifier = _name_estimators([self.estimator])[0][0]
colors = ['black', 'orange', 'blue', 'green', 'yellow', 'magenta'] * 2
linestyles = [':', '--', '-.', '-'] * 3
plt.figure()
for clf, label, clr, lsytle in zip(clf_list, clf_names, colors, linestyles):
mean_auc, mean_tpr, mean_fpr = self.get_auc(splitter, num_splits, clf)
legendString = "{}: Gini = {:.2f}".format(label, 2*mean_auc-1)
plt.plot(mean_fpr, mean_tpr, color=clr, linestyle=lsytle,
label=legendString, lw=2)
# plot the roc curve for ensemble
mean_auc, mean_tpr, mean_fpr = self.get_auc(splitter, num_splits, self.estimator)
ensemble_classifier = _name_estimators([self.estimator])[0][0]
legendString = "{}: Gini = {:.2f}".format(ensemble_classifier, 2*mean_auc-1)
plt.plot(mean_fpr, mean_tpr, color='red', linestyle='-',
label=legendString, lw=2)
# random line
plt.plot([0, 1], [0, 1], linestyle='--', color='gray', linewidth=2)
plt.xlim([-0.05, 1.05])
plt.ylim([-0.05, 1.05])
plt.xlabel('false positive rate')
plt.ylabel('true positive rate')
plt.title('Receiver Operator Characteristic')
plt.legend(loc="lower right")
plt.grid()
plt.tight_layout()
def __init__(self,
classifiers,
meta_classifier,
use_probas=False,
n_folds=2,
use_features_in_secondary=False,
stratify=True,
random_state=None,
shuffle=True,
verbose=0):
self.classifiers = classifiers
self.meta_classifier = meta_classifier
self.named_classifiers = {
key: value
for key, value in _name_estimators(classifiers)
}
self.named_meta_classifier = {
'meta-%s' % key: value
for key, value in _name_estimators([meta_classifier])
}
self.use_probas = use_probas
self.verbose = verbose
self.n_folds = n_folds
self.use_features_in_secondary = use_features_in_secondary
self.stratify = stratify
self.shuffle = shuffle
self.random_state = random_state
def fit(self, X, y):
"""Learn weight coefficients from training data for each classifier.
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.
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.weights and len(self.weights) != len(self.clfs):
raise ValueError('Number of classifiers and weights must be equal'
'; got %d weights, %d clfs' %
(len(self.weights), len(self.clfs)))
self.le_ = LabelEncoder()
self.le_.fit(y)
self.classes_ = self.le_.classes_
self.clfs_ = [clone(clf) for clf in self.clfs]
if self.verbose > 0:
print("Fitting %d classifiers..." % (len(self.clfs)))
for clf in self.clfs_:
if self.verbose > 0:
i = self.clfs_.index(clf) + 1
print("Fitting clf%d: %s (%d/%d)" %
(i, _name_estimators((clf, ))[0][0], i, len(self.clfs_)))
if self.verbose > 2:
if hasattr(clf, 'verbose'):
clf.set_params(verbose=self.verbose - 2)
if self.verbose > 1:
print(_name_estimators((clf, ))[0][1])
clf.fit(X, self.le_.transform(y))
return self
def fit(self, X, y):
"""Learn weight coefficients from training data for each classifier.
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.
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.weights and len(self.weights) != len(self.clfs):
raise ValueError('Number of classifiers and weights must be equal'
'; got %d weights, %d clfs'
% (len(self.weights), len(self.clfs)))
self.le_ = LabelEncoder()
self.le_.fit(y)
self.classes_ = self.le_.classes_
self.clfs_ = [clone(clf) for clf in self.clfs]
if self.verbose > 0:
print("Fitting %d classifiers..." % (len(self.clfs)))
for clf in self.clfs_:
if self.verbose > 0:
i = self.clfs_.index(clf) + 1
print("Fitting clf%d: %s (%d/%d)" %
(i, _name_estimators((clf,))[0][0], i, len(self.clfs_)))
if self.verbose > 2:
if hasattr(clf, 'verbose'):
clf.set_params(verbose=self.verbose - 2)
if self.verbose > 1:
print(_name_estimators((clf,))[0][1])
clf.fit(X, self.le_.transform(y))
return self
def __init__(self, regressors, meta_regressor, verbose=0):
self.regressors = regressors
self.meta_regressor = meta_regressor
self.named_regressors = {
key: value
for key, value in _name_estimators(regressors)
}
self.named_meta_regressor = {
'meta-%s' % key: value
for key, value in _name_estimators([meta_regressor])
}
self.verbose = verbose
def __init__( self , classifiers, class_labels = [], vote_method = "majority_vote", weights = None ):
"""
コンストラクタ（厳密にはイニシャライザ）
引数と同名のオブジェクトの属性を設定する必要あり（上位クラスの BaseEstimator仕様）
[Input]
classifiers_ : list
分類器のクラスのオブジェクトのリスト
"""
self.classifiers = classifiers
self.class_labels = class_labels
self.class_labels_ = class_labels
self.weights = weights
if classifiers != None:
self.__n_classifier = len( classifiers )
else:
self.__n_classifier = 0
self.vote_method = vote_method
# ?
if classifiers != None:
self.named_classifiers = { key: value
for key, value in _name_estimators(classifiers) }
else:
self.named_classifiers = {}
return
def __init__(self, classifiers, vote = 'classlabel', weights = None):
self.classifiers = classifiers
self.named_classifiers = { key:value for key, value
in _name_estimators(classifiers) }
self.vote = vote
self.weights = weights
def __init__(self, classifiers, vote='classlabel',
weights=None):
self.classifiers = classifiers
self.named_classifiers = {key: value for key, value in
_name_estimators(classifiers)}
self.vote = vote
self.weights = weights
def kfold_validate(self, estimator=None, score_method=None):
'''
K fold cross validation
Valid options are ['accuracy', 'adjusted_rand_score', 'average_precision',
'f1', 'f1_macro', 'f1_micro', 'f1_samples', 'f1_weighted', 'neg_log_loss',
'neg_mean_absolute_error', 'neg_mean_squared_error', 'neg_median_absolute_error',
'precision', 'precision_macro', 'precision_micro', 'precision_samples',
'precision_weighted', 'r2', 'recall', 'recall_macro', 'recall_micro',
'recall_samples', 'recall_weighted', 'roc_auc']
'''
if estimator is None:
estimator = self.estimator
if score_method is None:
score_method = self.score_method
estimator_name = _name_estimators([estimator])[0][0]
scores = cross_val_score(estimator=estimator,
X=self.X,
y=self.y,
cv=self.cv,
scoring = score_method,
n_jobs=self.n_jobs)
self._print_title(estimator_name)
print("{:d}-Fold {} score: {:.3f} +/- {:.3f}".format(self.cv, score_method,
np.mean(scores), np.std(scores)))
return (scores)
def make_pipeline(*steps):
"""Construct a Pipeline from the given estimators. This is a shorthand for
the Pipeline constructor; it does not require, and does not permit, naming
the estimators. Instead, their names will be set to the lowercase of their
types automatically.
Parameters
----------
*steps : list
List of estimators.
Returns
-------
p : Pipeline
Examples
--------
>>> from kenchi.outlier_detection import MiniBatchKMeans
>>> from kenchi.pipeline import make_pipeline
>>> from sklearn.preprocessing import StandardScaler
>>> scaler = StandardScaler()
>>> det = MiniBatchKMeans()
>>> pipeline = make_pipeline(scaler, det)
"""
return Pipeline(_name_estimators(steps))
def make_union(*transformers, **kwargs):
"""Construct a FeatureUnion from the given transformers.
This is a shorthand for the FeatureUnion constructor; it does not require,
and does not permit, naming the transformers. Instead, they will be given
names automatically based on their types. It also does not allow weighting.
Parameters
----------
*transformers : list of estimators
n_jobs : int or None, optional (default=None)
Number of jobs to run in parallel.
``None`` means 1 unless in a :obj:`joblib.parallel_backend` context.
``-1`` means using all processors. See :term:`Glossary <n_jobs>`
for more details.
Returns
-------
f : FeatureUnion
"""
n_jobs = kwargs.pop('n_jobs', None)
if kwargs:
# We do not currently support `transformer_weights` as we may want to
# change its type spec in make_union
raise TypeError('Unknown keyword arguments: "{}"'.format(
list(kwargs.keys())[0]))
return FeatureUnion(_name_estimators(transformers), n_jobs=n_jobs)
def make_pipeline(*steps):
"""Construct a Pipeline from the given estimators.
This is a shorthand for the Pipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Parameters
----------
*steps : list of estimators.
See Also
--------
sklearn.pipeline.Pipeline : Class for creating a pipeline of
transforms with a final estimator.
Examples
--------
>>> from sklearn.naive_bayes import GaussianNB
>>> from sklearn.preprocessing import StandardScaler
>>> make_pipeline(StandardScaler(), GaussianNB(priors=None))
Pipeline(steps=[('standardscaler', StandardScaler()),
('gaussiannb', GaussianNB())])
Returns
-------
p : Pipeline
"""
return Scaler(_name_estimators(steps))
def make_pipeline(*steps, **kwargs):
"""Construct a Pipeline from the given estimators.
This is a shorthand for the Pipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Parameters
----------
*steps : list of estimators.
memory : None, str or object with the joblib.Memory interface, optional
Used to cache the fitted transformers of the pipeline. By default,
no caching is performed. If a string is given, it is the path to
the caching directory. Enabling caching triggers a clone of
the transformers before fitting. Therefore, the transformer
instance given to the pipeline cannot be inspected
directly. Use the attribute ``named_steps`` or ``steps`` to
inspect estimators within the pipeline. Caching the
transformers is advantageous when fitting is time consuming.
Returns
-------
p : Pipeline
"""
memory = kwargs.pop('memory', None)
if kwargs:
raise TypeError('Unknown keyword arguments: "{}"'.format(
list(kwargs.keys())[0]))
return Pipeline(_name_estimators(steps), memory=memory)
def __init__(self, clfs, voting='hard', weights=None, verbose=0):
self.clfs = clfs
self.named_clfs = {key: value for key, value in _name_estimators(clfs)}
self.voting = voting
self.weights = weights
self.verbose = verbose
def __init__( self, regressors, weights = None, fitting = None, clone = False, debug = False ):
"""
Args :
regressors : list <regressors オブジェクト>
回帰器のオブジェクトのリスト
weights : list <float>
各回帰器の対する重みの値のリスト : __init()__ の引数と同名のオブジェクトの属性
fitting : list<bool>
各回帰器の対する学習を行うかのフラグのリスト
"""
self.regressors = regressors
self.fitted_regressors = regressors
self.weights = weights
self.clone = clone
self.debug = debug
if regressors != None:
self.n_classifier = len( regressors )
else:
self.n_classifier = 0
# regressors で指定した各オブジェクトの名前
if regressors != None:
self.named_regressors = { key: value for key, value in _name_estimators(regressors) }
else:
self.named_regressors = {}
if( self.debug ):
for i, named_classifier in enumerate(self.named_regressors):
print( "name {} : {}".format(i, self.named_regressors[named_classifier]) )
return
def __init__( self, regressors, final_regressors, second_regressors = None, n_splits = 4, clone = False, seed = 72 ):
self.regressors = regressors
self.fitted_regressors = regressors
self.final_regressors = final_regressors
self.second_regressors = second_regressors
self.fitted_second_regressors = second_regressors
self.n_classifier = len( regressors )
if( second_regressors != None ):
self.n_second_regressors = len( second_regressors )
else:
self.n_second_regressors = 0
self.n_splits = n_splits
self.clone = clone
self.seed = seed
self.accuracy = None
# classifiers で指定した各オブジェクトの名前
if regressors != None:
self.named_regressors = { key: value for key, value in _name_estimators(regressors) }
else:
self.named_regressors = {}
for i, named_regressor in enumerate(self.named_regressors):
print( "name {} : {}".format(i, self.named_regressors[named_regressor]) )
return
def __init__(self, clfs, voting='hard', weights=None, verbose=0):
self.clfs = clfs
self.named_clfs = {key: value for key, value in _name_estimators(clfs)}
self.voting = voting
self.weights = weights
self.verbose = verbose
def __init__(self, classifiers, vote='classlabel', weights=None):
self.classifiers = classifiers # list of classifiers
self.vote = vote # 'probability' or 'classlabel'
self.named_classifiers = {
key: value
for key, value in _name_estimators(classifiers)
}
self.weights = weights # weights for each of the classifiers
def __init__(self,
classifiers,
meta_classifier,
use_probas=False,
verbose=0):
self.classifiers = classifiers
self.meta_classifier = meta_classifier
self.named_classifiers = {
key: value
for key, value in _name_estimators(classifiers)
}
self.named_meta_classifier = {
'meta-%s' % key: value
for key, value in _name_estimators([meta_classifier])
}
self.use_probas = use_probas
self.verbose = verbose
def __init__(self, clfs, voting='hard', weights=None):
"""
voting: if 'hard', uses predicted class labels for majority rule voting
if 'soft', predicts the class label based on the argmax of the sums of the predicted probalities
"""
self.clfs = clfs
self.named_clfs = {key:value for key, value in _name_estimators(clfs)}
self.voting = voting
self.weights = weights
def __init__(self, clfs, voting, weights=None, threshold=None):
self.clfs = clfs
self.named_clfs = {key: value for key, value in _name_estimators(clfs)}
self.voting = voting
if voting is 'weighted':
self.combiner = WeightedVote(weights=weights, threshold=threshold)
elif voting is 'majority':
self.combiner = MajorityVote()
else:
raise AttributeError('Unrecognized voting method')
def __init__(self, clfs, voting='hard', weights=None):
"""
voting: if 'hard', uses predicted class labels for majority rule voting
if 'soft', predicts the class label based on the argmax of the sums of the predicted probalities
"""
self.clfs = clfs
# _name_estimators([LogisticRegression()]) ==> [('logisticregression', LogisticRegression(C=1.0, ....))]
self.named_clfs = {key:value for key, value in _name_estimators(clfs)}
self.voting = voting
self.weights = weights
def __init__(self, classifiers, vote='classlabel', weights=None):
self.classifiers = classifiers
# named_classifiers([..]) リストの中身は識別器で,
# この関数はそれを識別器の名前(sklearnが独自に決めた識別器のクラスの小文字)と，
# 識別器オブジェクトに分けられ，タプルのリストとなっている.
self.named_classifiers = {key: value for key, value
in _name_estimators(classifiers)}
self.vote = vote
self.weights = weights
def __init__(self, clfs, voting, weights=None, threshold=None):
self.clfs = clfs
self.named_clfs = {key:value for key,value in _name_estimators(clfs)}
self.voting=voting
if voting is 'weighted':
self.combiner=WeightedVote(weights=weights, threshold=threshold)
elif voting is 'majority':
self.combiner=MajorityVote()
else:
raise AttributeError('Unrecognized voting method')
def make_pipeline(*steps, **kwargs):
"""Construct a Pipeline from the given estimators.
This is a shorthand for the Pipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Parameters
----------
*steps : list of estimators.
memory : None, str or object with the joblib.Memory interface, optional
Used to cache the fitted transformers of the pipeline. By default,
no caching is performed. If a string is given, it is the path to
the caching directory. Enabling caching triggers a clone of
the transformers before fitting. Therefore, the transformer
instance given to the pipeline cannot be inspected
directly. Use the attribute ``named_steps`` or ``steps`` to
inspect estimators within the pipeline. Caching the
transformers is advantageous when fitting is time consuming.
verbose : boolean, optional (default=False)
If True, the time elapsed while fitting each step will be printed as it
is completed.
Returns
-------
p : Pipeline
See also
--------
imblearn.pipeline.Pipeline : Class for creating a pipeline of
transforms with a final estimator.
Examples
--------
>>> from sklearn.naive_bayes import GaussianNB
>>> from sklearn.preprocessing import StandardScaler
>>> make_pipeline(StandardScaler(), GaussianNB(priors=None))
... # doctest: +NORMALIZE_WHITESPACE
Pipeline(memory=None,
steps=[('standardscaler',
StandardScaler(copy=True, with_mean=True, with_std=True)),
('gaussiannb',
GaussianNB(priors=None, var_smoothing=1e-09))],
verbose=False)
"""
memory = kwargs.pop("memory", None)
verbose = kwargs.pop('verbose', False)
if kwargs:
raise TypeError('Unknown keyword arguments: "{}"'.format(
list(kwargs.keys())[0]))
return Pipeline(pipeline._name_estimators(steps),
memory=memory,
verbose=verbose)
def make_union(*transformers, **kwargs):
n_jobs = kwargs.pop('n_jobs', None)
verbose = kwargs.pop('verbose', False)
if kwargs:
# We do not currently support `transformer_weights` as we may want to
# change its type spec in make_union
raise TypeError('Unknown keyword arguments: "{}"'.format(
list(kwargs.keys())[0]))
return PandasFeatureUnion(_name_estimators(transformers),
n_jobs=n_jobs,
verbose=verbose)
def fit(self, X, y):
"""Learn weight coefficients from training data for each regressor.
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.
Returns
-------
self : object
"""
self.regr_ = [clone(regr) for regr in self.regressors]
self.meta_regr_ = clone(self.meta_regressor)
if self.verbose > 0:
print("Fitting %d regressors..." % (len(self.regressors)))
for regr in self.regr_:
if self.verbose > 0:
i = self.regr_.index(regr) + 1
print("Fitting regressor%d: %s (%d/%d)" %
(i, _name_estimators(
(regr, ))[0][0], i, len(self.regr_)))
if self.verbose > 2:
if hasattr(regr, 'verbose'):
regr.set_params(verbose=self.verbose - 2)
if self.verbose > 1:
print(_name_estimators((regr, ))[0][1])
regr.fit(X, y)
meta_features = self._predict_meta_features(X)
self.meta_regr_.fit(meta_features, y)
return self
def _get_transformer_list(estimators):
"""
Construct (name, trans, column) tuples from list
"""
message = ('`make_column_transformer` expects (transformer, columns)')
transformers, columns = zip(*estimators)
names, _ = zip(*_name_estimators(transformers))
transformer_list = list(zip(names, transformers, columns))
return transformer_list
def make_pipeline(*steps):
"""Construct a Pipeline from the given estimators.
This is a shorthand for the Pipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Returns
-------
p : Pipeline
"""
return Pipeline(pipeline._name_estimators(steps))
def make_pipeline(*steps):
"""Construct a Pipeline from the given estimators.
This is a shorthand for the Pipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Returns
-------
p : Pipeline
"""
return Pipeline(pipeline._name_estimators(steps))
def get_all_performance_metrics(self, metrics=None):
"get metrics for all classifiers"
(clf_list, clf_names) = self.get_component_classifiers()
ensemble_classifier = _name_estimators([self.estimator])[0][0]
for clf, label in zip(clf_list, clf_names):
print("\n{}\n{}\n{}\n".format("*"*60, label, "*"*60))
self.get_performance_metrics(metrics, clf)
print("\n{}\n{}\n{}\n".format("*"*60, ensemble_classifier, "*"*60))
self.get_performance_metrics(metrics, self.estimator)
def fit(self, X, y):
""" Fit ensemble classifers and the meta-classifier.
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.
Returns
-------
self : object
"""
self.clfs_ = [clone(clf) for clf in self.classifiers]
self.meta_clf_ = clone(self.meta_classifier)
if self.verbose > 0:
print("Fitting %d classifiers..." % (len(self.classifiers)))
for clf in self.clfs_:
if self.verbose > 0:
i = self.clfs_.index(clf) + 1
print("Fitting classifier%d: %s (%d/%d)" %
(i, _name_estimators((clf, ))[0][0], i, len(self.clf_)))
if self.verbose > 2:
if hasattr(clf, 'verbose'):
clf.set_params(verbose=self.verbose - 2)
if self.verbose > 1:
print(_name_estimators((clf, ))[0][1])
clf.fit(X, y)
meta_features = self._predict_meta_features(X)
self.meta_clf_.fit(meta_features, y)
return self
def __init__(self,
classifiers,
vote='probability',
weights=None,
method='majority_voting'):
self.classifiers = classifiers
self.named_classifiers = {
k: v
for k, v in _name_estimators(classifiers)
}
self.vote = vote
self.weights = weights
self.method = method
def __init__(self,
classifiers,
weights=None,
fitting=None,
vote_method="majority_vote"):
"""
Args :
classifiers : list <classifier オブジェクト>
分類器のクラスのオブジェクトのリスト
weights : list <float>
各分類器の対する重みの値のリスト : __init()__ の引数と同名のオブジェクトの属性
fitting : list<bool>
各分類器の対する学習を行うかのフラグのリスト
vote_method : str ( "majority_vote" or "probability_vote" )
アンサンブルによる最終的な判断判断手法 : __init()__ の引数と同名のオブジェクトの属性
"majority_vote" : 弱識別器の多数決で決定する.多数決方式 (＝クラスラベルの argmax() 結果）
"probability_vote" : 弱識別器の重み付け結果で決定する.（＝クラスの所属確率の argmax() 結果）
"""
self.classifiers = classifiers
self.fitting = fitting
self.fitted_classifiers = classifiers
self.weights = weights
self.n_classes = 0
if classifiers != None:
self.n_classifier = len(classifiers)
else:
self.n_classifier = 0
self.vote_method = vote_method
self.encoder = LabelEncoder()
# classifiers で指定した各オブジェクトの名前
if classifiers != None:
self.named_classifiers = {
key: value
for key, value in _name_estimators(classifiers)
}
else:
self.named_classifiers = {}
for i, named_classifier in enumerate(self.named_classifiers):
print("name {} : {}".format(
i, self.named_classifiers[named_classifier]))
if fitting == None:
fitting = []
for i in range(len(self.classifiers)):
fitting.append(True)
return
def __init__(self, classifiers, vote='classlabel', weights=None):
"""
Constructor
"""
self.classifiers = classifiers
self.named_classifiers = {
key: value
for key, value in _name_estimators(classifiers)
}
self.vote = vote
self.weights = weights
self.lablenc_ = LabelEncoder()
self.classifiers_ = []
self.classes_ = []
def __init__(
self,
estimator=None,
output_dtype=float,
output_dims=((None, np.nan),),
fit_input="data",
transform_input="data",
estimator_name=None,
model_path=None,
features_dir=None,
extension=".hdf5",
save_func=None,
load_func=None,
dataset_map=None,
input_dask_array=False,
fit_kwargs=None,
**kwargs,
):
super().__init__(**kwargs)
self.estimator = estimator
self.output_dtype = output_dtype
if not all(len(d) == 2 for d in output_dims):
raise ValueError(
"output_dims must be an iterable of size 2 tuples "
f"(dim_name, dim_size), not {output_dims}"
)
self.output_dims = output_dims
self.fit_input = fit_input
self.transform_input = transform_input
if estimator_name is None:
estimator_name = _name_estimators([estimator])[0][0]
self.estimator_name = estimator_name
self.model_path = model_path
self.features_dir = features_dir
self.extension = extension
estimator_save_fn = (
None
if estimator is None
else estimator._get_tags().get("bob_features_save_fn")
)
estimator_load_fn = (
None
if estimator is None
else estimator._get_tags().get("bob_features_load_fn")
)
self.save_func = save_func or estimator_save_fn or save
self.load_func = load_func or estimator_load_fn or load
self.dataset_map = dataset_map
self.input_dask_array = input_dask_array
self.fit_kwargs = fit_kwargs or {}
def make_debug_pipeline(*steps, **kwargs):
"""Construct a DebugPipeline from the given estimators.
This is a shorthand for the DebugPipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Parameters
----------
*steps : list of estimators.
memory : None, str or object with the joblib.Memory interface, optional
Used to cache the fitted transformers of the pipeline. By default,
no caching is performed. If a string is given, it is the path to
the caching directory. Enabling caching triggers a clone of
the transformers before fitting. Therefore, the transformer
instance given to the pipeline cannot be inspected
directly. Use the attribute ``named_steps`` or ``steps`` to
inspect estimators within the pipeline. Caching the
transformers is advantageous when fitting is time consuming.
verbose : boolean, default=False
If True, the time elapsed while fitting each step will be printed as it
is completed.
log_callback: string, default=None.
The callback function that logs information in between each
intermediate step. Defaults to None. If set to `'default'`,
:func:`default_log_callback` is used.
See :func:`default_log_callback` for an example.
See Also
--------
sklego.pipeline.DebugPipeline : Class for creating a pipeline of
transforms with a final estimator.
Examples
--------
>>> from sklearn.naive_bayes import GaussianNB
>>> from sklearn.preprocessing import StandardScaler
>>> make_debug_pipeline(StandardScaler(), GaussianNB(priors=None))
DebugPipeline(steps=[('standardscaler', StandardScaler()),
('gaussiannb', GaussianNB())])
Returns
-------
p : DebugPipeline
"""
memory = kwargs.pop('memory', None)
verbose = kwargs.pop('verbose', False)
log_callback = kwargs.pop('log_callback', None)
if kwargs:
raise TypeError('Unknown keyword arguments: "{}"'
.format(list(kwargs.keys())[0]))
return DebugPipeline(_name_estimators(steps), memory=memory, verbose=verbose, log_callback=log_callback)
def make_pipeline(*steps, **kwargs):
"""Construct a Pipeline from the given estimators.
This is a shorthand for the Pipeline constructor; it does not require, and
does not permit, naming the estimators. Instead, their names will be set
to the lowercase of their types automatically.
Parameters
----------
*steps : list of estimators.
memory : None, str or object with the joblib.Memory interface, optional
Used to cache the fitted transformers of the pipeline. By default,
no caching is performed. If a string is given, it is the path to
the caching directory. Enabling caching triggers a clone of
the transformers before fitting. Therefore, the transformer
instance given to the pipeline cannot be inspected
directly. Use the attribute ``named_steps`` or ``steps`` to
inspect estimators within the pipeline. Caching the
transformers is advantageous when fitting is time consuming.
Returns
-------
p : Pipeline
See also
--------
imblearn.pipeline.Pipeline : Class for creating a pipeline of
transforms with a final estimator.
Examples
--------
>>> from sklearn.naive_bayes import GaussianNB
>>> from sklearn.preprocessing import StandardScaler
>>> make_pipeline(StandardScaler(), GaussianNB(priors=None))
... # doctest: +NORMALIZE_WHITESPACE
Pipeline(memory=None,
steps=[('standardscaler',
StandardScaler(copy=True, with_mean=True, with_std=True)),
('gaussiannb',
GaussianNB(priors=None, var_smoothing=1e-09))])
"""
memory = kwargs.pop('memory', None)
if kwargs:
raise TypeError('Unknown keyword arguments: "{}"'
.format(list(kwargs.keys())[0]))
return Pipeline(pipeline._name_estimators(steps), memory=memory)
def make_sparkunion(*transformers):
"""Construct a FeatureUnion from the given transformers.
This is a shorthand for the FeatureUnion constructor; it does not require,
and does not permit, naming the transformers. Instead, they will be given
names automatically based on their types. It also does not allow weighting.
Examples
--------
>>> from sklearn.decomposition import PCA, TruncatedSVD
>>> make_union(PCA(), TruncatedSVD()) # doctest: +NORMALIZE_WHITESPACE
FeatureUnion(n_jobs=1,
transformer_list=[('pca', PCA(copy=True, n_components=None,
whiten=False)),
('truncatedsvd',
TruncatedSVD(algorithm='randomized',
n_components=2, n_iter=5,
random_state=None, tol=0.0))],
transformer_weights=None)
Returns
-------
f : FeatureUnion
"""
return SparkFeatureUnion(_name_estimators(transformers))
def __init__(self, clfs, voting="hard", weights=None):
self.clfs = clfs
self.named_clfs = {key: value for key, value in _name_estimators(clfs)}
self.voting = voting
self.weights = weights
def transform(self, X, y=None):
xform_data = self.transform_.transform(X, y)
return np.append(X, xform_data, axis=1)
class LogExpPipeline(Pipeline):
def fit(self, X, y):
super(LogExpPipeline, self).fit(X, np.log1p(y))
def predict(self, X):
return np.expm1(super(LogExpPipeline, self).predict(X))
#
# Model/pipeline with scaling,pca,svm
# knn
knn_pipe = LogExpPipeline(_name_estimators([RobustScaler(),
KNeighborsRegressor(n_neighbors = 15, metric = 'cityblock')]))
#
svm_pipe = LogExpPipeline(_name_estimators([RobustScaler(),
SVR(kernel='rbf', C=30, epsilon=0.05)]))
# results = cross_val_score(svm_pipe, train, y_train, cv=5, scoring='r2')
# print("SVM score: %.4f (%.4f)" % (results.mean(), results.std()))
# exit()
#
# Model/pipeline with scaling,pca,ElasticNet
#
en = ElasticNet(alpha=0.01, l1_ratio=0.9)
#
# XGBoost model
def transform(self, X, y=None):
xform_data = self.transform_.transform(X, y)
return np.append(X, xform_data, axis=1)
class LogExpPipeline(Pipeline):
def fit(self, X, y):
super(LogExpPipeline, self).fit(X, y)
def predict(self, X):
return super(LogExpPipeline, self).predict(X)
#
# Model/pipeline with scaling,pca,svm
# knn
knn_pipe = LogExpPipeline(_name_estimators([RobustScaler(),
KNeighborsClassifier(n_neighbors = 15, metric = 'cityblock')]))
#
svm_pipe = LogExpPipeline(_name_estimators([RobustScaler(),
SVC(kernel='rbf', C=14)]))
# results = cross_val_score(svm_pipe, train, y_train, cv=5, scoring='r2')
# print("SVM score: %.4f (%.4f)" % (results.mean(), results.std()))
# exit()
#
# XGBoost model
#
xgb_model = xgb.XGBClassifier(max_depth=4, learning_rate=0.0045, subsample=0.921,nthread=6,
objective='multi:softmax', n_estimators=500)
def __init__(self, classifiers):
self.classifiers = classifiers
self.named_classifiers = {key: value for key, value in _name_estimators(classifiers)}
def make_alpha_pipeline(*steps):
return AlphaPipeline(_name_estimators(steps))
def make_dataframe_pipeline(steps):
"""Construct a DataFramePipeline from the given estimators."""
return DataFramePipeline(_name_estimators(steps))
def make_transformer_pipeline(*steps):
"""Construct a TransformerPipeline from the given estimators.
"""
return TransformerPipeline(_name_estimators(steps))