def _fit_resample(self, X, y):
n_samples = X.shape[0]
# convert y to z_score
y_z = (y - y.mean()) / y.std()
index0 = np.arange(n_samples)
index_negative = index0[y_z > self.negative_thres]
index_positive = index0[y_z <= self.positive_thres]
index_unclassified = [x for x in index0
if x not in index_negative
and x not in index_positive]
y_z[index_negative] = 0
y_z[index_positive] = 1
y_z[index_unclassified] = -1
ros = RandomOverSampler(
sampling_strategy=self.sampling_strategy,
random_state=self.random_state,
ratio=self.ratio)
_, _ = ros.fit_resample(X, y_z)
sample_indices = ros.sample_indices_
print("Before sampler: %s. Total after: %s"
% (Counter(y_z), sample_indices.shape))
self.sample_indices_ = np.array(sample_indices)
if self.return_indices:
return (_safe_indexing(X, sample_indices),
_safe_indexing(y, sample_indices),
sample_indices)
return (_safe_indexing(X, sample_indices),
_safe_indexing(y, sample_indices))
def train_submission(self, module_path, X, y, train_idx=None):
"""Train the estimator of a given submission.
Parameters
----------
module_path : str
The path to the submission where `filename` is located.
X : {array-like, sparse matrix, dataframe} of shape \
(n_samples, n_features)
The data matrix.
y : array-like of shape (n_samples,)
The target vector.
train_idx : array-like of shape (n_training_samples,), default=None
The training indices. By default, the full dataset will be used
to train the model. If an array is provided, `X` and `y` will be
subsampled using these indices.
Returns
-------
estimator : estimator object
The scikit-learn fitted on (`X`, `y`).
"""
train_idx = slice(None, None, None) if train_idx is None else train_idx
submission_module = import_module_from_source(
os.path.join(module_path, self.filename),
os.path.splitext(self.filename)[0], # keep the module name only
sanitize=True)
estimator = submission_module.get_estimator()
X_train = _safe_indexing(X, train_idx)
y_train = _safe_indexing(y, train_idx)
return estimator.fit(X_train, y_train)
def test_check_fit_params(indices):
X = np.random.randn(4, 2)
fit_params = {
'list': [1, 2, 3, 4],
'array': np.array([1, 2, 3, 4]),
'sparse-col': sp.csc_matrix([1, 2, 3, 4]).T,
'sparse-row': sp.csc_matrix([1, 2, 3, 4]),
'scalar-int': 1,
'scalar-str': 'xxx',
'None': None,
}
result = _check_fit_params(X, fit_params, indices)
indices_ = indices if indices is not None else list(range(X.shape[0]))
for key in ['sparse-row', 'scalar-int', 'scalar-str', 'None']:
assert result[key] is fit_params[key]
assert result['list'] == _safe_indexing(fit_params['list'], indices_)
assert_array_equal(
result['array'], _safe_indexing(fit_params['array'], indices_)
)
assert_allclose_dense_sparse(
result['sparse-col'],
_safe_indexing(fit_params['sparse-col'], indices_)
)
def test_safe_indexing_2d_container_axis_1(array_type, indices_type, indices):
# validation of the indices
# we make a copy because indices is mutable and shared between tests
indices_converted = copy(indices)
if indices_type == "slice" and isinstance(indices[1], int):
indices_converted[1] += 1
columns_name = ["col_0", "col_1", "col_2"]
array = _convert_container(
[[1, 2, 3], [4, 5, 6], [7, 8, 9]], array_type, columns_name
)
indices_converted = _convert_container(indices_converted, indices_type)
if isinstance(indices[0], str) and array_type != "dataframe":
err_msg = (
"Specifying the columns using strings is only supported "
"for pandas DataFrames"
)
with pytest.raises(ValueError, match=err_msg):
_safe_indexing(array, indices_converted, axis=1)
else:
subset = _safe_indexing(array, indices_converted, axis=1)
assert_allclose_dense_sparse(
subset, _convert_container([[2, 3], [5, 6], [8, 9]], array_type)
)
def __getitem__(self, index):
X_resampled = _safe_indexing(
self.X,
self.indices_[
index * self.batch_size:(index + 1) * self.batch_size
],
)
y_resampled = _safe_indexing(
self.y,
self.indices_[
index * self.batch_size:(index + 1) * self.batch_size
],
)
if issparse(X_resampled) and not self.keep_sparse:
X_resampled = X_resampled.toarray()
if self.sample_weight is not None:
sample_weight_resampled = _safe_indexing(
self.sample_weight,
self.indices_[
index * self.batch_size:(index + 1) * self.batch_size
],
)
if self.sample_weight is None:
return X_resampled, y_resampled
else:
return X_resampled, y_resampled, sample_weight_resampled
def _fit_resample(self, X, y):
random_state = check_random_state(self.random_state)
idx_under = np.empty((0, ), dtype=int)
for target_class in np.unique(y):
if target_class in self.sampling_strategy_.keys():
n_samples = self.sampling_strategy_[target_class]
index_target_class = random_state.choice(
range(np.count_nonzero(y == target_class)),
size=n_samples,
replace=self.replacement,
)
else:
index_target_class = slice(None)
idx_under = np.concatenate(
(
idx_under,
np.flatnonzero(y == target_class)[index_target_class],
),
axis=0,
)
self.sample_indices_ = idx_under
return _safe_indexing(X, idx_under), _safe_indexing(y, idx_under)
def _fit_and_predict_oof_model(
self,
estimator: RegressorMixin,
X: ArrayLike,
y: ArrayLike,
train_index: ArrayLike,
val_index: ArrayLike,
sample_weight: Optional[ArrayLike] = None,
) -> Tuple[RegressorMixin, NDArray, ArrayLike]:
"""
Fit a single out-of-fold model on a given training set and
perform predictions on a test set.
Parameters
----------
estimator : RegressorMixin
Estimator to train.
X : ArrayLike of shape (n_samples, n_features)
Input data.
y : ArrayLike of shape (n_samples,)
Input labels.
train_index : ArrayLike of shape (n_samples_train)
Training data indices.
val_index : ArrayLike of shape (n_samples_val)
Validation data indices.
sample_weight : Optional[ArrayLike] of shape (n_samples,)
Sample weights. If None, then samples are equally weighted.
By default ``None``.
Returns
-------
Tuple[RegressorMixin, NDArray, ArrayLike]
- [0]: RegressorMixin, fitted estimator
- [1]: NDArray of shape (n_samples_val,),
estimator predictions on the validation fold.
- [3]: ArrayLike of shape (n_samples_val,),
validation data indices.
"""
X_train = _safe_indexing(X, train_index)
y_train = _safe_indexing(y, train_index)
X_val = _safe_indexing(X, val_index)
if sample_weight is None:
estimator = fit_estimator(estimator, X_train, y_train)
else:
sample_weight_train = _safe_indexing(sample_weight, train_index)
estimator = fit_estimator(
estimator, X_train, y_train, sample_weight_train
)
if _num_samples(X_val) > 0:
y_pred = estimator.predict(X_val)
else:
y_pred = np.array([])
return estimator, y_pred, val_index
def _split_fit_score_trial(self, X, y, idx=0):
"""
Splits the dataset, fits a clone of the estimator, then scores it
according to the required metrics.
The index of the split is added to the random_state if the
random_state is not None; this ensures that every split is shuffled
differently but in a deterministic fashion for testing purposes.
"""
random_state = self.random_state
if random_state is not None:
random_state += idx
splitter = self._check_cv(self.cv, random_state)
for train_index, test_index in splitter.split(X, y):
# Safe indexing handles multiple types of inputs including
# DataFrames and structured arrays - required for generic splits.
X_train = _safe_indexing(X, train_index)
y_train = _safe_indexing(y, train_index)
X_test = _safe_indexing(X, test_index)
y_test = _safe_indexing(y, test_index)
model = clone(self.estimator)
model.fit(X_train, y_train)
if hasattr(model, "predict_proba"):
# Get the probabilities for the positive class
y_scores = model.predict_proba(X_test)[:, 1]
else:
# Use the decision function to get the scores
y_scores = model.decision_function(X_test)
# Compute the curve metrics and thresholds
curve_metrics = precision_recall_curve(y_test, y_scores)
precision, recall, thresholds = curve_metrics
# Compute the F1 score from precision and recall
# Don't need to warn for F, precision/recall would have warned
with np.errstate(divide="ignore", invalid="ignore"):
beta = self.fbeta**2
f_score = (1 + beta) * precision * recall / (beta * precision +
recall)
# Ensure thresholds ends at 1
thresholds = np.append(thresholds, 1)
# Compute the queue rate
queue_rate = np.array([(y_scores >= threshold).mean()
for threshold in thresholds])
yield {
"thresholds": thresholds,
"precision": precision,
"recall": recall,
"fscore": f_score,
"queue_rate": queue_rate,
}
def _fit_resample(self, X, y):
self._validate_estimator()
enn = EditedNearestNeighbours(
sampling_strategy=self.sampling_strategy,
n_neighbors=self.n_neighbors,
kind_sel="mode",
n_jobs=self.n_jobs,
)
enn.fit_resample(X, y)
index_not_a1 = enn.sample_indices_
index_a1 = np.ones(y.shape, dtype=bool)
index_a1[index_not_a1] = False
index_a1 = np.flatnonzero(index_a1)
# clean the neighborhood
target_stats = Counter(y)
class_minority = min(target_stats, key=target_stats.get)
# compute which classes to consider for cleaning for the A2 group
classes_under_sample = [
c
for c, n_samples in target_stats.items()
if (
c in self.sampling_strategy_.keys()
and (n_samples > X.shape[0] * self.threshold_cleaning)
)
]
self.nn_.fit(X)
class_minority_indices = np.flatnonzero(y == class_minority)
X_class = _safe_indexing(X, class_minority_indices)
y_class = _safe_indexing(y, class_minority_indices)
nnhood_idx = self.nn_.kneighbors(X_class, return_distance=False)[:, 1:]
nnhood_label = y[nnhood_idx]
if self.kind_sel == "mode":
nnhood_label_majority, _ = mode(nnhood_label, axis=1)
nnhood_bool = np.ravel(nnhood_label_majority) == y_class
elif self.kind_sel == "all":
nnhood_label_majority = nnhood_label == class_minority
nnhood_bool = np.all(nnhood_label, axis=1)
else:
raise NotImplementedError
# compute a2 group
index_a2 = np.ravel(nnhood_idx[~nnhood_bool])
index_a2 = np.unique(
[index for index in index_a2 if y[index] in classes_under_sample]
)
union_a1_a2 = np.union1d(index_a1, index_a2).astype(int)
selected_samples = np.ones(y.shape, dtype=bool)
selected_samples[union_a1_a2] = False
self.sample_indices_ = np.flatnonzero(selected_samples)
return (
_safe_indexing(X, self.sample_indices_),
_safe_indexing(y, self.sample_indices_),
)
def check_null_weight(
sample_weight: Optional[ArrayLike], X: ArrayLike,
y: ArrayLike) -> Tuple[Optional[NDArray], ArrayLike, ArrayLike]:
"""
Check sample weights and remove samples with null sample weights.
Parameters
----------
sample_weight : Optional[ArrayLike] of shape (n_samples,)
Sample weights.
X : ArrayLike of shape (n_samples, n_features)
Training samples.
y : ArrayLike of shape (n_samples,)
Training labels.
Returns
-------
sample_weight : Optional[NDArray] of shape (n_samples,)
Non-null sample weights.
X : ArrayLike of shape (n_samples, n_features)
Training samples with non-null weights.
y : ArrayLike of shape (n_samples,)
Training labels with non-null weights.
Examples
--------
>>> import numpy as np
>>> from mapie.utils import check_null_weight
>>> X = np.array([[0], [1], [2], [3], [4], [5]])
>>> y = np.array([5, 7, 9, 11, 13, 15])
>>> sample_weight = np.array([0, 1, 1, 1, 1, 1])
>>> sample_weight, X, y = check_null_weight(sample_weight, X, y)
>>> print(sample_weight)
[1. 1. 1. 1. 1.]
>>> print(X)
[[1]
[2]
[3]
[4]
[5]]
>>> print(y)
[ 7 9 11 13 15]
"""
if sample_weight is not None:
sample_weight = _check_sample_weight(sample_weight, X)
non_null_weight = sample_weight != 0
X = _safe_indexing(X, non_null_weight)
y = _safe_indexing(y, non_null_weight)
sample_weight = _safe_indexing(sample_weight, non_null_weight)
sample_weight = cast(Optional[NDArray], sample_weight)
return sample_weight, X, y
def _fit_resample(self, X, y):
# Find the nearest neighbour of every point
nn = NearestNeighbors(n_neighbors=2, n_jobs=self.n_jobs)
nn.fit(X)
nns = nn.kneighbors(X, return_distance=False)[:, 1]
links = self.is_tomek(y, nns, self.sampling_strategy_)
self.sample_indices_ = np.flatnonzero(np.logical_not(links))
return (
_safe_indexing(X, self.sample_indices_),
_safe_indexing(y, self.sample_indices_),
)
def test_safe_indexing_1d_array_error(X_constructor):
# check that we are raising an error if the array-like passed is 1D and
# we try to index on the 2nd dimension
X = list(range(5))
if X_constructor == 'array':
X_constructor = np.asarray(X)
elif X_constructor == 'series':
pd = pytest.importorskip("pandas")
X_constructor = pd.Series(X)
err_msg = "'X' should be a 2D NumPy array, 2D sparse matrix or pandas"
with pytest.raises(ValueError, match=err_msg):
_safe_indexing(X_constructor, [0, 1], axis=1)
def generator(X, y, sample_weight, indices, batch_size):
while True:
for index in range(0, len(indices), batch_size):
X_res = _safe_indexing(X, indices[index:index + batch_size])
y_res = _safe_indexing(y, indices[index:index + batch_size])
if issparse(X_res) and not keep_sparse:
X_res = X_res.toarray()
if sample_weight is None:
yield X_res, y_res
else:
sw_res = _safe_indexing(sample_weight,
indices[index:index + batch_size])
yield X_res, y_res, sw_res
def accumulated_local_effects(est, x, feature, n_quantiles):
"""calculates ale for a feature"""
ale = np.array(n_quantiles,)
features_indices = np.asarray(
_get_column_indices(x, feature), dtype=np.int32, order='C'
).ravel()
quantiles = _quantiles_from_x(_safe_indexing(x, features_indices, axis=1), n_quantiles)
x_feat = _safe_indexing(x, feature, axis=1)
if x_feat.to_numpy().dtype.name == "category" or x_feat.to_numpy().dtype == "object":
ale = _ale_for_categorical(est, quantiles, x, x_feat)
else:
ale = _ale_for_numeric(est, quantiles, x, x_feat)
return ale
def _fit_resample(self, X, y, sample_weight=None):
self._validate_estimator()
random_state = check_random_state(self.random_state)
target_stats = Counter(y)
class_minority = min(target_stats, key=target_stats.get)
idx_under = np.empty((0, ), dtype=int)
for target_class in np.unique(y):
if target_class in self.sampling_strategy_.keys():
# select a sample from the current class
idx_maj = np.flatnonzero(y == target_class)
sel_idx_maj = random_state.randint(
low=0,
high=target_stats[target_class],
size=self.n_seeds_S)
idx_maj_sample = idx_maj[sel_idx_maj]
minority_class_indices = np.flatnonzero(y == class_minority)
C_indices = np.append(minority_class_indices, idx_maj_sample)
# create the set composed of all minority samples and one
# sample from the current class.
C_x = _safe_indexing(X, C_indices)
C_y = _safe_indexing(y, C_indices)
# create the set S with removing the seed from S
# since that it will be added anyway
idx_maj_extracted = np.delete(idx_maj, sel_idx_maj, axis=0)
S_x = _safe_indexing(X, idx_maj_extracted)
S_y = _safe_indexing(y, idx_maj_extracted)
self.estimator_.fit(C_x, C_y)
pred_S_y = self.estimator_.predict(S_x)
S_misclassified_indices = np.flatnonzero(pred_S_y != S_y)
idx_tmp = idx_maj_extracted[S_misclassified_indices]
idx_under = np.concatenate(
(idx_under, idx_maj_sample, idx_tmp), axis=0)
else:
idx_under = np.concatenate(
(idx_under, np.flatnonzero(y == target_class)), axis=0)
X_resampled = _safe_indexing(X, idx_under)
y_resampled = _safe_indexing(y, idx_under)
# apply Tomek cleaning
tl = TomekLinks(sampling_strategy=list(self.sampling_strategy_.keys()))
X_cleaned, y_cleaned = tl.fit_resample(X_resampled, y_resampled)
self.sample_indices_ = _safe_indexing(idx_under, tl.sample_indices_)
idx_under = self.sample_indices_
if sample_weight is not None:
# sample_weight is already validated in self.fit_resample()
sample_weight_under = _safe_indexing(sample_weight, idx_under)
return X_cleaned, y_cleaned, sample_weight_under
else:
return X_cleaned, y_cleaned
def evaluate(D, sol):
phenotype = SamplingBenchmark.map_to_phenotype(
CustomSamplingBenchmark.to_phenotype(sol))
X_sampled = _safe_indexing(self.X_train, phenotype)
y_sampled = _safe_indexing(self.y_train, phenotype)
if X_sampled.shape[0] > 0:
cls = self.evaluator.fit(X_sampled, y_sampled)
y_predicted = cls.predict(self.X_valid)
quality = accuracy_score(self.y_valid, y_predicted)
size_percentage = len(y_sampled) / len(sol)
return (1 - quality) * size_percentage
else:
return math.inf
def _fit_transformer(self, y):
"""Check transformer and fit transformer.
Create the default transformer, fit it and make additional inverse
check on a subset (optional).
"""
if (self.transformer is not None and
(self.func is not None or self.inverse_func is not None)):
raise ValueError("'transformer' and functions 'func'/"
"'inverse_func' cannot both be set.")
elif self.transformer is not None:
self.transformer_ = clone(self.transformer)
else:
if self.func is not None and self.inverse_func is None:
raise ValueError("When 'func' is provided, 'inverse_func' must"
" also be provided")
self.transformer_ = FunctionTransformer(
func=self.func, inverse_func=self.inverse_func, validate=True,
check_inverse=self.check_inverse)
# XXX: sample_weight is not currently passed to the
# transformer. However, if transformer starts using sample_weight, the
# code should be modified accordingly. At the time to consider the
# sample_prop feature, it is also a good use case to be considered.
self.transformer_.fit(y)
if self.check_inverse:
idx_selected = slice(None, None, max(1, y.shape[0] // 10))
y_sel = _safe_indexing(y, idx_selected)
y_sel_t = self.transformer_.transform(y_sel)
if not np.allclose(y_sel,
self.transformer_.inverse_transform(y_sel_t)):
warnings.warn("The provided functions or transformer are"
" not strictly inverse of each other. If"
" you are sure you want to proceed regardless"
", set 'check_inverse=False'", UserWarning)
def _fit_resample(self, X, y):
self._validate_estimator()
X_resampled = [X.copy()]
y_resampled = [y.copy()]
for class_sample, n_samples in self.sampling_strategy_.items():
print(self.sampling_strategy_.items())
if n_samples == 0:
continue
target_class_indices = np.flatnonzero(y == class_sample)
X_class = _safe_indexing(X, target_class_indices)
self.nn_k_.fit(X_class)
nns = self.nn_k_.kneighbors(X_class, return_distance=False)[:, 1:]
X_new, y_new = self._make_samples(X_class, y.dtype, class_sample,
X_class, nns, n_samples, 1.0)
X_resampled.append(X_new)
y_resampled.append(y_new)
if sparse.issparse(X):
X_resampled = sparse.vstack(X_resampled, format=X.format)
else:
X_resampled = np.vstack(X_resampled)
y_resampled = np.hstack(y_resampled)
from IPython.display import display, HTML
import pandas as pd
pd.set_option("display.max_rows", None, "display.max_columns", None)
new = pd.DataFrame(X_new)
display(new)
return X_resampled, y_resampled
def _local_parallel_build_trees(sampler,
tree,
forest,
X,
y,
sample_weight,
tree_idx,
n_trees,
verbose=0,
class_weight=None,
n_samples_bootstrap=None):
# resample before to fit the tree
X_resampled, y_resampled = sampler.fit_resample(X, y)
if sample_weight is not None:
sample_weight = _safe_indexing(sample_weight, sampler.sample_indices_)
if _get_n_samples_bootstrap is not None:
n_samples_bootstrap = min(n_samples_bootstrap, X_resampled.shape[0])
tree = _parallel_build_trees(
tree,
forest,
X_resampled,
y_resampled,
sample_weight,
tree_idx,
n_trees,
verbose=verbose,
class_weight=class_weight,
n_samples_bootstrap=n_samples_bootstrap,
)
return sampler, tree
def make_sample(imbalanced_data_arr2, diff):
# 将数据集分开为少数类数据和多数类数据
minor_data_arr2, major_data_arr2 = seperate_minor_and_major_data(imbalanced_data_arr2)
imbalanced_featured_data = imbalanced_data_arr2[:, : -1]
imbalanced_label_data = imbalanced_data_arr2[:, -1]
# 原始少数样本的特征集
old_feature_data = minor_data_arr2[:, : -1]
# 原始少数样本的标签值
old_label_data = minor_data_arr2[0][-1]
danger_index = in_danger(imbalanced_featured_data, old_feature_data, old_label_data, imbalanced_label_data)
# 少数样本中噪音集合，也就是最终要产生新样本的集合
danger_index_data = _safe_indexing(old_feature_data, danger_index)
# 获取每一个少数类样本点周围最近的n_neighbors-1个点的位置矩阵
nns = NearestNeighbors(n_neighbors=6).fit(old_feature_data).kneighbors(danger_index_data,
return_distance=False)[:, 1:]
# 随机产生diff个随机数作为之后产生新样本的选取的样本下标值
samples_indices = np.random.randint(low=0, high=np.shape(danger_index_data)[0], size=diff)
# 随机产生diff个随机数作为之后产生新样本的间距值
steps = np.random.uniform(size=diff)
cols = np.mod(samples_indices, nns.shape[1])
reshaped_feature = np.zeros((diff, danger_index_data.shape[1]))
for i, (col, step) in enumerate(zip(cols, steps)):
row = samples_indices[i]
reshaped_feature[i] = danger_index_data[row] - step * (danger_index_data[row] - old_feature_data[nns[row, col]])
new_min_feature_data = np.vstack((reshaped_feature, old_feature_data))
return new_min_feature_data
def test_safe_indexing_1d_container_mask(array_type, indices_type):
indices = [False] + [True] * 2 + [False] * 6
array = _convert_container([1, 2, 3, 4, 5, 6, 7, 8, 9], array_type)
indices = _convert_container(indices, indices_type)
subset = _safe_indexing(array, indices, axis=0)
assert_allclose_dense_sparse(subset, _convert_container([2, 3],
array_type))
def _init_pretest(self, features, target):
"""Set the sample of data used to verify pipelines work
with the passed data set.
This is not intend for anything other than perfunctory dataset
pipeline compatibility testing
"""
num_unique_target = len(np.unique(target))
# make sure train_size is at least num_unique_target
train_size = max(min(50, int(0.9 * features.shape[0])),
num_unique_target)
self.pretest_X, _, self.pretest_y, _ = \
train_test_split(
features,
target,
random_state=self.random_state,
test_size=None,
train_size=train_size
)
#Make sure there is a least one example from each class
#for this evaluative test sample
if not np.array_equal(np.unique(target), np.unique(self.pretest_y)):
unique_target_idx = np.unique(target, return_index=True)[1]
self.pretest_y[0:unique_target_idx.shape[0]] = \
_safe_indexing(target, unique_target_idx)
def _fit_resample(self, X, y):
self._validate_estimator()
X_resampled = [X.copy()]
y_resampled = [y.copy()]
for class_sample, n_samples in self.sampling_strategy_.items():
if n_samples == 0:
continue
target_class_indices = np.flatnonzero(y == class_sample)
X_class = _safe_indexing(X, target_class_indices)
self.nn_k_.fit(X_class)
nns = self.nn_k_.kneighbors(X_class, return_distance=False)[:, 1:]
X_new, y_new = self._make_samples(
X_class, y.dtype, class_sample, X_class, nns, n_samples, 1.0
)
X_resampled.append(X_new)
y_resampled.append(y_new)
if sparse.issparse(X):
X_resampled = sparse.vstack(X_resampled, format=X.format)
else:
X_resampled = np.vstack(X_resampled)
y_resampled = np.hstack(y_resampled)
return X_resampled, y_resampled
def calculate(self, chromosome: ndarray) -> float:
labels = self.cluster.run(chromosome, self.samples)
self.samples, labels = check_X_y(self.samples, labels)
le = LabelEncoder()
labels = le.fit_transform(labels)
n_samples, _ = self.samples.shape
n_labels = len(le.classes_)
check_number_of_labels(n_labels, n_samples)
intra_dists = np.zeros(n_labels)
centroids = np.zeros((n_labels, len(self.samples[0])), dtype=float)
for k in range(n_labels):
cluster_k = _safe_indexing(self.samples, labels == k)
centroid = chromosome[k]
centroids[k] = centroid
intra_dists[k] = np.average(
pairwise_distances(cluster_k, [centroid]))
centroid_distances = pairwise_distances(centroids)
if np.allclose(intra_dists, 0) or np.allclose(centroid_distances, 0):
return 0.0
centroid_distances[centroid_distances == 0] = np.inf
combined_intra_dists = intra_dists[:, None] + intra_dists
scores = np.max(combined_intra_dists / centroid_distances, axis=1)
return 1 / np.mean(scores)
def _boost_real(self, iboost, X, y, sample_weight, random_state):
"""Implement a single boost using the SAMME.R real algorithm."""
estimator, sampler = self._make_sampler_estimator(random_state=random_state)
X_res, y_res = sampler.fit_resample(X, y)
sample_weight_res = _safe_indexing(sample_weight, sampler.sample_indices_)
estimator.fit(X_res, y_res, sample_weight=sample_weight_res)
y_predict_proba = estimator.predict_proba(X)
if iboost == 0:
self.classes_ = getattr(estimator, "classes_", None)
self.n_classes_ = len(self.classes_)
y_predict = self.classes_.take(np.argmax(y_predict_proba, axis=1), axis=0)
# Instances incorrectly classified
incorrect = y_predict != y
# Error fraction
estimator_error = np.mean(np.average(incorrect, weights=sample_weight, axis=0))
# Stop if classification is perfect
if estimator_error <= 0:
return sample_weight, 1.0, 0.0
# Construct y coding as described in Zhu et al [2]:
#
# y_k = 1 if c == k else -1 / (K - 1)
#
# where K == n_classes_ and c, k in [0, K) are indices along the second
# axis of the y coding with c being the index corresponding to the true
# class label.
n_classes = self.n_classes_
classes = self.classes_
y_codes = np.array([-1.0 / (n_classes - 1), 1.0])
y_coding = y_codes.take(classes == y[:, np.newaxis])
# Displace zero probabilities so the log is defined.
# Also fix negative elements which may occur with
# negative sample weights.
proba = y_predict_proba # alias for readability
np.clip(proba, np.finfo(proba.dtype).eps, None, out=proba)
# Boost weight using multi-class AdaBoost SAMME.R alg
estimator_weight = (
-1.0
* self.learning_rate
* ((n_classes - 1.0) / n_classes)
* (y_coding * np.log(y_predict_proba)).sum(axis=1)
)
# Only boost the weights if it will fit again
if not iboost == self.n_estimators - 1:
# Only boost positive weights
sample_weight *= np.exp(
estimator_weight * ((sample_weight > 0) | (estimator_weight < 0))
)
return sample_weight, 1.0, estimator_error
def permutations(estimator,
X,
y,
cv=None,
n_permuations=100,
random_state=0,
scoring=None):
"""
This follows the sklearn API sklearn.inspection.permutation_test_score
I have modified accordinlgy to accomodate filtering of features using correlation matrix
before running cross-validation using the model
"""
Xs, ys = indexable(X, y)
cv = check_cv(cv, y, classifier=is_classifier(estimator))
scorer = check_scoring(estimator, scoring=scoring)
random_state = check_random_state(random_state)
# corr = CorrMatrix()
# corr.fit(X,y)
# Xs, ys = corr.transform()
score = _permutations(clone(estimator), Xs, ys, cv, scorer)
permutation_scores = np.zeros((n_permuations))
for i in range(n_permuations):
# corr_p = CorrMatrix()
# corr_p.fit(X, y)
# Xp, yp = corr_p.transform()
yp = _safe_indexing(y, random_state.permutation(len(y)))
permutation_scores[i] = _permutations(clone(estimator), Xs, yp, cv,
scorer)
pvalue = (np.sum(permutation_scores >= score) + 1.0) / (n_permuations + 1)
return score, permutation_scores, pvalue
def _fit_resample(self, X, y):
self._validate_estimator()
X_resampled = X.copy()
y_resampled = y.copy()
for class_sample, n_samples in self.sampling_strategy_.items():
if n_samples == 0:
continue
target_class_indices = np.flatnonzero(y == class_sample)
X_class = _safe_indexing(X, target_class_indices)
self.nn_k_.fit(X_class)
nns = self.nn_k_.kneighbors(X_class, return_distance=False)[:, 1:]
X_new, y_new = self._make_samples(X_class, y.dtype, class_sample,
X_class, nns, n_samples, 1.0)
if sparse.issparse(X_new):
X_resampled = sparse.vstack([X_resampled, X_new])
sparse_func = "tocsc" if X.format == "csc" else "tocsr"
X_resampled = getattr(X_resampled, sparse_func)()
else:
X_resampled = np.vstack((X_resampled, X_new))
y_resampled = np.hstack((y_resampled, y_new))
return X_resampled, y_resampled
def db(X, labels): X, labels = check_X_y(X, labels) le = LabelEncoder() labels = le.fit_transform(labels) n_samples, _ = X.shape n_labels = len(le.classes_) check_number_of_labels(n_labels, n_samples) intra_dists = np.zeros(n_labels) centroids = np.zeros((n_labels, len(X[0])), dtype=float) for k in range(n_labels): cluster_k = _safe_indexing(X, labels == k) centroid = cluster_k.mean(axis=0) centroids[k] = centroid intra_dists[k] = np.average(pairwise_distances( cluster_k, [centroid], metric='euclidean')) centroid_distances = pairwise_distances(centroids, metric='euclidean') if np.allclose(intra_dists, 0) or np.allclose(centroid_distances, 0): return 0.0 centroid_distances[centroid_distances == 0] = np.inf combined_intra_dists = intra_dists[:, None] + intra_dists scores = np.max(combined_intra_dists / centroid_distances, axis=1) return np.mean(scores)
def SMOTE_Borderline_D(imbalanced_data_arr2):
# 将数据集分开为少数类数据和多数类数据
minor_data_arr2, major_data_arr2 = seperate_minor_and_major_data(
imbalanced_data_arr2)
print('多数类样本数：', len(major_data_arr2), ', 少数类样本数：', len(minor_data_arr2))
imbalanced_featured_data = imbalanced_data_arr2[:, :-1]
imbalanced_label_data = imbalanced_data_arr2[:, -1]
# 计算多数类数据和少数类数据之间的数量差,也是需要过采样的数量
# 1
n = major_data_arr2.shape[0] - minor_data_arr2.shape[0]
# 原始少数样本的特征集
old_feature_data = minor_data_arr2[:, :-1]
# 原始少数样本的标签值
old_label_data = minor_data_arr2[0][-1]
danger_index = in_danger(imbalanced_featured_data, old_feature_data,
old_label_data, imbalanced_label_data)
# 少数样本中噪音集合，也就是最终要产生新样本的集合
danger_index_data = _safe_indexing(old_feature_data, danger_index)
# 使用K近邻方法产生的新样本特征集
new_feature_data = make_sample(old_feature_data, danger_index_data,
n) # 扩展少数类样本集
# 将类别标签数组合并到少数类样本特征集，构建出新的少数类样本数据集
new_labels_data = np.array([old_label_data] * len(new_feature_data))
new_minor_data_arr2 = np.column_stack((new_feature_data, new_labels_data))
# balanced_data_arr2 = np.row_stack((new_minor_data_arr2, major_data_arr2))
balanced_data_arr2 = np.row_stack((major_data_arr2, new_minor_data_arr2))
# 将少数类数据集和多数据类数据集合并，并对样本数据进行打乱重排，
# balanced_data_arr2 = concat_and_shuffle_data(new_minor_data_arr2, major_data_arr2)
return balanced_data_arr2
def _fit_resample(self, X, y):
# FIXME: to be removed in 0.12
if self.n_jobs is not None:
warnings.warn(
"The parameter `n_jobs` has been deprecated in 0.10 and will be "
"removed in 0.12. You can pass an nearest neighbors estimator where "
"`n_jobs` is already set instead.",
FutureWarning,
)
self._validate_estimator()
X_resampled = [X.copy()]
y_resampled = [y.copy()]
for class_sample, n_samples in self.sampling_strategy_.items():
if n_samples == 0:
continue
target_class_indices = np.flatnonzero(y == class_sample)
X_class = _safe_indexing(X, target_class_indices)
self.nn_k_.fit(X_class)
nns = self.nn_k_.kneighbors(X_class, return_distance=False)[:, 1:]
X_new, y_new = self._make_samples(X_class, y.dtype, class_sample,
X_class, nns, n_samples, 1.0)
X_resampled.append(X_new)
y_resampled.append(y_new)
if sparse.issparse(X):
X_resampled = sparse.vstack(X_resampled, format=X.format)
else:
X_resampled = np.vstack(X_resampled)
y_resampled = np.hstack(y_resampled)
return X_resampled, y_resampled
