def test_surf_pandas_inputs():
"""Check: Data (pandas DataFrame/Series): SURF works with pandas DataFrame and Series inputs"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2),
RandomForestClassifier(n_estimators=100, n_jobs=-1))
assert np.mean(cross_val_score(clf, features_df, labels_s, cv=3,
n_jobs=-1)) > 0.7
def test_surf_pipeline_parallel():
"""Check: Data (Binary Endpoint, Discrete Features): SURF works in a sklearn pipeline when SURF is parallelized"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2, n_jobs=-1),
RandomForestClassifier(n_estimators=100, n_jobs=-1))
assert np.mean(cross_val_score(clf, features, labels, cv=3)) > 0.7
def test_surf_pipeline():
"""Ensure that SURF works in a sklearn pipeline when it is parallelized"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2, n_jobs=-1),
RandomForestClassifier(n_estimators=100, n_jobs=-1))
assert np.mean(cross_val_score(clf, features, labels, cv=3)) > 0.7
def test_surf_pipeline_mixed_attributes():
"""Check: Data (Mixed Attributes): SURF works in a sklearn pipeline"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2),
RandomForestClassifier(n_estimators=100, n_jobs=-1))
assert np.mean(cross_val_score(clf, features_mixed_attributes,
labels_mixed_attributes, cv=3, n_jobs=-1)) > 0.7
def test_surf_pipeline_cont_endpoint():
"""Check: Data (Continuous Endpoint): SURF works in a sklearn pipeline"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2),
RandomForestRegressor(n_estimators=100, n_jobs=-1))
assert abs(np.mean(cross_val_score(clf, features_cont_endpoint,
labels_cont_endpoint, cv=3, n_jobs=-1))) < 0.5
def test_surf_pipeline_missing_values():
"""Ensure that SURF works in a sklearn pipeline with missing values"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2, n_jobs=-1), Imputer(),
RandomForestClassifier(n_estimators=100, n_jobs=-1))
assert np.mean(
cross_val_score(
clf, features_missing_values, labels_missing_values, cv=3)) > 0.7
def test_surf_init():
"""Check: SURF, SURF*, and MultiSURF constructors store custom values correctly"""
clf = SURF(n_features_to_select=7,
discrete_threshold=20,
verbose=True,
n_jobs=3)
assert clf.n_features_to_select == 7
assert clf.discrete_threshold == 20
assert clf.verbose == True
assert clf.n_jobs == 3
def test_surf_pipeline_multiclass():
"""Check: Data (Multiclass Endpoint): SURF works in a sklearn pipeline"""
np.random.seed(240932)
clf = make_pipeline(SURF(n_features_to_select=2), SimpleImputer(),
RandomForestClassifier(n_estimators=100, n_jobs=-1))
assert np.mean(
cross_val_score(
clf, features_multiclass, labels_multiclass, cv=3,
n_jobs=-1)) > 0.7
def get_selector(name, estimator=None, n_features_to_select=None, **params):
if name == 'RobustSelector':
return RobustSelector(estimator, n_features_to_select=n_features_to_select, **search_dict(params,
('cv', 'verbose')))
elif name == 'MaxFeatures':
return SelectFromModel(estimator, threshold=-np.inf, max_features=n_features_to_select)
elif name == 'RandomSubsetSelector':
return RandomSubsetSelector(estimator, n_features_to_select=n_features_to_select, **search_dict(params,
('n_subsets', 'subset_size', 'random_state')))
elif name == 'FeatureImportanceThreshold':
return SelectFromModel(estimator, **search_dict(params, 'threshold'))
elif name == 'RFE':
return RFE(estimator, n_features_to_select=n_features_to_select, **search_dict(params,
('step', 'verbose')))
elif name == 'RFECV':
return RFECV(estimator, n_features_to_select=n_features_to_select, **search_dict(params,
('step', 'cv', 'verbose')))
elif name == 'FoldChangeFilter':
return FoldChangeFilter(**search_dict(params,
('threshold', 'direction', 'below', 'pseudo_count')))
elif name == 'ZeroFractionFilter':
return ZeroFractionFilter(**search_dict(params,
('threshold',)))
elif name == 'RpkmFilter':
return RpkmFilter(**search_dict(params,
('threshold',)))
elif name == 'RpmFilter':
return RpmFilter(**search_dict(params,
('threshold',)))
elif name == 'DiffExpFilter':
return DiffExpFilter(max_features=n_features_to_select, **search_dict(params,
('threshold', 'script', 'temp_dir', 'score_type', 'method')))
elif name == 'ReliefF':
from skrebate import ReliefF
return ReliefF(n_features_to_select=n_features_to_select,
**search_dict(params, ('n_jobs', 'n_neighbors', 'discrete_limit')))
elif name == 'SURF':
from skrebate import SURF
return SURF(n_features_to_select=n_features_to_select,
**search_dict(params, ('n_jobs', 'discrete_limit')))
elif name == 'MultiSURF':
from skrebate import MultiSURF
return MultiSURF(n_features_to_select=n_features_to_select,
**search_dict(params, ('n_jobs', 'discrete_limit')))
elif name == 'SIS':
return SIS(n_features_to_select=n_features_to_select,
**search_dict(params, ('temp_dir', 'sis_params')))
elif name == 'NullSelector':
return NullSelector()
else:
raise ValueError('unknown selector: {}'.format(name))
GradientBoostingClassifier(random_state=args.randomseed),
'rf':
RandomForestClassifier(n_jobs=-1, random_state=args.randomseed),
'ext':
ExtraTreesClassifier(n_estimators=100,
n_jobs=-1,
random_state=args.randomseed),
'svm':
SVC(kernel='sigmoid', random_state=args.randomseed)
}
# 初始化 fs 字典
fs = {
'ReliefF': ReliefF(n_features_to_select=100, verbose=False, n_jobs=-1),
# 'TuRF': TuRF(core_algorithm="ReliefF", n_features_to_select=100, verbose=False, n_jobs=-1),
'SURF': SURF(n_features_to_select=100, verbose=False, n_jobs=-1),
'SURFstar': SURFstar(n_features_to_select=100,
verbose=False,
n_jobs=-1)
}
print('\nClassifier parameters:', clf[args.classifier])
print('\nStarting cross validating without feature selection...\n')
# 特征排序前的增量特征预测，根据名称调用字典中指定分类器
y_pred_list = [
cross_val_predict(clf[args.classifier],
X[:, 0:i + 1],
y,
cv=args.kfolds,
def relief(X, y):
np.random.seed(0)
return SURF().fit(X, y).feature_importances_
def relief(X, y):
return SURF().fit(X, y).feature_importances_
def FeatureWeights(self, weights=("pearson", "variance"), **kwargs):
"""
Calculates the requested weights and log them
:param weights: a list of weights, a subset of {'pearson', 'variance', 'relieff',
'surf', 'sobol', 'morris', 'delta_mmnt', 'info-gain'}
:param kwargs: all input acceptable by ``skrebate.ReliefF``, ``skrebate.surf``,
``sensapprx.SensAprx``
:return: None
"""
from pandas import DataFrame, read_sql
self.data = read_sql("SELECT * FROM data", self.conn)
features = list(self.data.columns)
features.remove(self.target)
weights_df = read_sql("SELECT * FROM weights", self.conn)
if len(weights_df) == 0:
weights_df = DataFrame({"feature": features})
X = self.data[features].values
y = self.data[self.target].values
n_features = kwargs.get("n_features", int(len(features) / 2))
domain = None
probs = None
regressor = kwargs.get("regressor", None)
reduce = kwargs.get("reduce", True)
num_smpl = kwargs.get("num_smpl", 700)
W = {"feature": features}
for factor in weights:
if factor == "pearson":
Res = dict(self.data.corr(method="pearson").fillna(0)[self.target])
W["pearson"] = [Res[v] for v in features]
elif factor == "variance":
Res = dict(self.data.var())
W["variance"] = [Res[v] for v in features]
elif factor == "relieff":
from skrebate import ReliefF
n_neighbors = kwargs.get("n_neighbors", 80)
RF = ReliefF(n_features_to_select=n_features, n_neighbors=n_neighbors)
RF.fit(X, y)
W["relieff"] = [
RF.feature_importances_[features.index(v)] for v in features
]
elif factor == "surf":
from skrebate import SURF
RF = SURF(n_features_to_select=n_features)
RF.fit(X, y)
W["surf"] = [
RF.feature_importances_[features.index(v)] for v in features
]
elif factor == "sobol":
from .sensapprx import SensAprx
SF = SensAprx(
method="sobol",
domain=domain,
probs=probs,
regressor=regressor,
reduce=reduce,
num_smpl=num_smpl,
)
SF.fit(X, y)
domain = SF.domain
probs = SF.probs
W["sobol"] = [SF.weights_[features.index(v)] for v in features]
elif factor == "morris":
from .sensapprx import SensAprx
SF = SensAprx(
method="morris",
domain=domain,
probs=probs,
regressor=regressor,
reduce=reduce,
num_smpl=num_smpl,
)
SF.fit(X, y)
domain = SF.domain
probs = SF.probs
W["morris"] = [SF.weights_[features.index(v)] for v in features]
elif factor == "delta-mmnt":
from .sensapprx import SensAprx
SF = SensAprx(
method="delta-mmnt",
domain=domain,
probs=probs,
regressor=regressor,
reduce=reduce,
num_smpl=num_smpl,
)
SF.fit(X, y)
domain = SF.domain
probs = SF.probs
W["delta_mmnt"] = [SF.weights_[features.index(v)] for v in features]
elif factor == "info-gain":
from sklearn.feature_selection import mutual_info_classif
Res = mutual_info_classif(X, y, discrete_features=True)
W["info_gain"] = [Res[features.index(v)] for v in features]
new_w_df = DataFrame(W)
merged = weights_df.merge(new_w_df, on="feature")
merged.fillna(0.0)
merged.to_sql("weights", self.conn, if_exists="replace", index=False)
def rank_features_by_rebate_methods(data_split_list,
fs_method,
iterate,
remove_percent=0.1,
verbose=False):
## 0. Input arguments:
# data_split_list: data frame that contains the learning data
# fs_method: feature ranking methods to be used: 'SURF', 'SURFstar', 'MultiSURF', or 'MultiSURFstar'
# iterate: whether to implement TURF: True or False (TURF will remove low-ranking features after each iteration, effective when #features is large)
# remove_percent: percentage of features removed at each iteration (only applied when iterate = True)
# verbose: whether to show progress by each fold: True of False
## 1. Define function for feature ranking method
# SURF
if fs_method == 'SURF':
# Implement TURF extension when 'iterate == True'
if iterate == True:
fs = TuRF(core_algorithm='SURF', pct=remove_percent)
else:
fs = SURF()
# SURFstar
if fs_method == 'SURFstar':
if iterate == True:
fs = TuRF(core_algorithm='SURFstar', pct=remove_percent)
else:
fs = SURFstar()
# MultiSURF
if fs_method == 'MultiSURF':
if iterate == True:
fs = TuRF(core_algorithm='MultiSURF', pct=remove_percent)
else:
fs = MultiSURF()
# MultiSURFstar
if fs_method == 'MultiSURFstar':
if iterate == True:
fs = TuRF(core_algorithm='MultiSURFstar', pct=remove_percent)
else:
fs = MultiSURFstar()
## 2. Perform feature ranking on each fold of training data
# iterate by folds
feat_impt_dict = {}
for i in range(0, len(data_split_list)):
# intermediate output
if verbose == True:
print('Computing feature importance scores using data from fold ' +
str(i) + '\n')
# obtain training feature matrix and response vector
feat_train, label_train, _, _ = data_split_list[i]
# fit feature ranking model using the specified method
if iterate == True:
fs.fit(feat_train.values, label_train.values, list(feat_train))
else:
fs.fit(feat_train.values, label_train.values)
# output feature importance scores in a data frame
fold_name = 'Fold_' + str(i)
feat_impt_dict[fold_name] = fs.feature_importances_
# aggregate results from muliple folds into one data frame
feat_impt_df = pd.DataFrame(feat_impt_dict)
feat_impt_df.index = feat_train.columns
return feat_impt_df
