Geometric smote

Class to perform over-sampling using Geometric SMOTE.

GeometricSMOTE(sampling_strategy='auto', k_neighbors=5, truncation_factor=1.0, deformation_factor=0.0, selection_strategy='combined', categorical_features=None, random_state=None)

Bases: BaseOverSampler

Class to to perform over-sampling using Geometric SMOTE.

This algorithm is an implementation of Geometric SMOTE, a geometrically enhanced drop-in replacement for SMOTE. Read more in the [user_guide].

Parameters:

Name	Type	Description	Default
categorical_features	ArrayLike | None	Specified which features are categorical. Can either be: - array of indices specifying the categorical features. - mask array of shape (n_features, ) and `bool` dtype for which `True` indicates the categorical features.	None
sampling_strategy	dict[int, int] | str | float | Callable	Sampling information to resample the data set. When float, it corresponds to the desired ratio of the number of samples in the minority class over the number of samples in the majority class after resampling. It is only available for binary classification. When str, specify the class targeted by the resampling. The number of samples in the different classes will be equalized. Possible choices are: 'minority': resample only the minority class. 'not minority': resample all classes but the minority class. 'not majority': resample all classes but the majority class. 'all': resample all classes. 'auto': equivalent to 'not majority'. When dict, the keys correspond to the targeted classes. The values correspond to the desired number of samples for each targeted class. When callable, function taking y and returns a dict. The keys correspond to the targeted classes. The values correspond to the desired number of samples for each class.	'auto'
random_state	RandomState | int | None	Control the randomization of the algorithm. If int, it is the seed used by the random number generator. If np.random.RandomState instance, it is the random number generator. If None, the random number generator is the RandomState instance used by np.random.	None
truncation_factor	float	The type of truncation. The values should be in the [-1.0, 1.0] range.	1.0
deformation_factor	float	The type of geometry. The values should be in the [0.0, 1.0] range.	0.0
selection_strategy	str	The type of Geometric SMOTE algorithm with the following options: 'combined', 'majority', 'minority'.	'combined'
k_neighbors	NearestNeighbors | int	If int, number of nearest neighbours to use when synthetic samples are constructed for the minority method. If object, an estimator that inherits from sklearn.neighbors.base.KNeighborsMixin class that will be used to find the k_neighbors.	5

Attributes:

Name	Type	Description
n_features_in_		int Number of features in the input dataset.
nns_pos_		estimator object Validated k-nearest neighbours created from the k_neighbors parameter. It is used to find the nearest neighbors of the same class of a selected observation.
nn_neg_		estimator object Validated k-nearest neighbours created from the k_neighbors parameter. It is used to find the nearest neighbor of the remaining classes (k=1) of a selected observation.
random_state_	RandomState	An instance of np.random.RandomState class.
sampling_strategy_	dict[int, int]	Actual sampling strategy.

Examples:

>>> import numpy as np
>>> from collections import Counter
>>> from sklearn.datasets import make_classification
>>> from imblearn_extra.gsmote import GeometricSMOTE
>>> np.set_printoptions(legacy='1.25')
>>> X, y = make_classification(n_classes=2, class_sep=2,
... weights=[0.1, 0.9], n_informative=3, n_redundant=1, flip_y=0,
... n_features=20, n_clusters_per_class=1, n_samples=1000, random_state=10)
>>> print('Original dataset shape %s' % Counter(y))
Original dataset shape Counter({{1: 900, 0: 100}})
>>> gsmote = GeometricSMOTE(random_state=1)
>>> X_resampled, y_resampled = gsmote.fit_resample(X, y)
>>> print('Resampled dataset shape %s' % Counter(y_resampled))
Resampled dataset shape Counter({{0: 900, 1: 900}})

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def __init__(
    self: Self,
    sampling_strategy: dict[int, int] | str | float | Callable = 'auto',
    k_neighbors: NearestNeighbors | int = 5,
    truncation_factor: float = 1.0,
    deformation_factor: float = 0.0,
    selection_strategy: str = 'combined',
    categorical_features: ArrayLike | None = None,
    random_state: np.random.RandomState | int | None = None,
) -> None:
    """Initialize oversampler."""
    super().__init__(sampling_strategy=sampling_strategy)
    self.k_neighbors = k_neighbors
    self.truncation_factor = truncation_factor
    self.deformation_factor = deformation_factor
    self.selection_strategy = selection_strategy
    self.categorical_features = categorical_features
    self.random_state = random_state

_check_X_y(X, y)

Check input and output data.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _check_X_y(  # noqa: N802
    self: Self,
    X: ArrayLike | sparse.csc_matrix | sparse.csr_matrix,
    y: ArrayLike,
) -> tuple[NDArray, NDArray, bool]:
    """Check input and output data."""
    y, binarize_y = check_target_type(y, indicate_one_vs_all=True)
    validated_data: tuple[NDArray | sparse.csc_matrix | sparse.csr_matrix, NDArray] = validate_data(
        self,
        X,
        y,
        reset=True,
        dtype=None,
        accept_sparse=['csr', 'csc'],
    )
    X, y = validated_data
    if not isinstance(X, np.ndarray):
        X = X.toarray()
    return X, y, binarize_y

_decode_categorical(X_init, X_resampled)

Reverses the encoding of the categorical features.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _decode_categorical(self: Self, X_init: NDArray, X_resampled: NDArray) -> NDArray:
    """Reverses the encoding of the categorical features."""

    if self.categorical_features is None:
        return X_resampled.astype(X_init.dtype)

    if math.isclose(self.median_std_, 0):
        X_resampled[: X_init.shape[0], self.continuous_features_.size :] = self.ohe_.transform(
            X_init[:, self.categorical_features_],
        ).toarray()

    indices_reordered = np.argsort(
        np.hstack((self.continuous_features_, self.categorical_features_)),
    )
    X_resampled = np.hstack(
        (
            X_resampled[:, : self.continuous_features_.size],
            self.ohe_.inverse_transform(X_resampled[:, self.continuous_features_.size :]),
        ),
    )[:, indices_reordered].astype(X_init.dtype)
    return X_resampled

_encode_categorical(X, y)

Encode categorical features.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _encode_categorical(self: Self, X: NDArray, y: NDArray) -> NDArray:
    """Encode categorical features."""

    if self.categorical_features is None:
        return X

    # Compute the median of the standard deviation of the minority class
    class_minority = Counter(y).most_common()[-1][0]

    # Calcuate variance
    X_continuous = check_array(X[:, self.continuous_features_], dtype=np.float64)
    X_minority_continuous = X_continuous[np.flatnonzero(y == class_minority)]
    var = X_minority_continuous.var(axis=0)
    self.median_std_ = np.median(np.sqrt(var))

    # OneHotEncoder
    X_categorical = X[:, self.categorical_features_]
    X_ohe_categorical = self.ohe_.transform(X_categorical)
    X_ohe_categorical.data = (
        np.ones_like(X_ohe_categorical.data, dtype=X_ohe_categorical.dtype) * self.median_std_ / 2
    )
    X_encoded = np.hstack([X_continuous, X_ohe_categorical.toarray()])

    return X_encoded

_make_geometric_samples(X_init, X, y, pos_class_label, n_samples)

A support function that returns artificials samples.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _make_geometric_samples(  # noqa: C901
    self: Self,
    X_init: NDArray,
    X: NDArray,
    y: NDArray,
    pos_class_label: str | int,
    n_samples: int,
) -> tuple[NDArray, NDArray, list[NDArray]]:
    """A support function that returns artificials samples."""

    # Return zero new samples
    if n_samples == 0:
        return (
            np.array([], dtype=X.dtype).reshape(0, X.shape[1]),
            np.array([], dtype=y.dtype),
            [],
        )

    # Select positive class samples
    X_pos = X[y == pos_class_label]

    # Force minority strategy if no negative class samples are present
    self.selection_strategy_ = 'minority' if X.shape[0] == X_pos.shape[0] else self.selection_strategy

    # Minority or combined strategy
    if self.selection_strategy_ in ('minority', 'combined'):
        self.nns_pos_.fit(X_pos)
        points_pos = self.nns_pos_.kneighbors(X_pos)[1][:, 1:]
        samples_indices = self.random_state_.randint(
            low=0,
            high=len(points_pos.flatten()),
            size=n_samples,
        )
        rows = np.floor_divide(samples_indices, points_pos.shape[1])
        cols = np.mod(samples_indices, points_pos.shape[1])

    # Majority or combined strategy
    if self.selection_strategy_ in ('majority', 'combined'):
        X_neg = X[y != pos_class_label]
        self.nn_neg_.fit(X_neg)
        points_neg = self.nn_neg_.kneighbors(X_pos)[1]
        if self.selection_strategy_ == 'majority':
            samples_indices = self.random_state_.randint(
                low=0,
                high=len(points_neg.flatten()),
                size=n_samples,
            )
            rows = np.floor_divide(samples_indices, points_neg.shape[1])
            cols = np.mod(samples_indices, points_neg.shape[1])

    # Case that the median std equals to zeros
    if self.categorical_features is not None and math.isclose(self.median_std_, 0):
        X[:, self.continuous_features_.size :] = self.ohe_.transform(
            X_init[:, self.categorical_features_],
        ).toarray()
        X_pos = X[y == pos_class_label]
        if self.selection_strategy_ in ('majority', 'combined'):
            X_neg = X[y != pos_class_label]

    # Generate new samples
    X_new = np.zeros((n_samples, X.shape[1]))
    all_neighbors = []
    for ind, (row, col) in enumerate(zip(rows, cols, strict=False)):
        # Define center point
        center = X_pos[row]

        # Minority strategy
        if self.selection_strategy_ == 'minority':
            surface_point = X_pos[points_pos[row, col]]
            neighbors = X_pos[points_pos[row]]

        # Majority strategy
        elif self.selection_strategy_ == 'majority':
            surface_point = X_neg[points_neg[row, col]]
            neighbors = X_neg[points_neg[row]]

        # Combined strategy
        else:
            surface_point_pos = X_pos[points_pos[row, col]]
            surface_point_neg = X_neg[points_neg[row, 0]]
            radius_pos = norm(center - surface_point_pos)
            radius_neg = norm(center - surface_point_neg)
            surface_point = surface_point_neg if radius_pos > radius_neg else surface_point_pos
            neighbors = np.vstack([X_pos[points_pos[row]], X_neg[points_neg[row]]])

        if self.categorical_features is not None:
            all_neighbors.append(neighbors)

        # Append new sample - no categorical features
        X_new[ind] = make_geometric_sample(
            center,
            surface_point,
            self.truncation_factor,
            self.deformation_factor,
            self.random_state_,
        )

    # Create new samples for target variable
    y_new = np.array([pos_class_label] * len(samples_indices))

    return X_new, y_new, all_neighbors

_populate_categorical_features(X_new, y_new, all_neighbors)

A support function that populates categorical features.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _populate_categorical_features(
    self: Self,
    X_new: NDArray,
    y_new: NDArray,
    all_neighbors: list[NDArray],
) -> tuple[NDArray, NDArray]:
    """A support function that populates categorical features."""
    categories_size = (
        [self.continuous_features_.size] + [cat.size for cat in self.ohe_.categories_]
        if self.categorical_features is not None
        else None
    )
    for ind, neighbors in enumerate(all_neighbors):
        X_new[ind] = populate_categorical_features(
            X_new[ind],
            neighbors,
            categories_size,
            self.random_state_,
        )
    return X_new, y_new

_validate_categorical_features()

Validate categorical features.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _validate_categorical_features(self: Self) -> Self:
    """Validate categorical features."""

    if self.categorical_features is None:
        self.categorical_features_ = np.flatnonzero([])
        self.continuous_features_: NDArray = np.arange(self.n_features_in_)
        return self

    categorical_features = np.asarray(self.categorical_features)
    if categorical_features.dtype.name == 'bool':
        self.categorical_features_ = np.flatnonzero(categorical_features)
    else:
        if any(index not in np.arange(self.n_features_in_) for index in categorical_features):
            error_msg = (
                'Some of the categorical indices are out of range. Indices'
                f' should be between 0 and {self.n_features_in_ - 1}.'
            )
            raise ValueError(error_msg)
        self.categorical_features_ = np.sort(categorical_features)
    self.continuous_features_ = np.setdiff1d(
        np.arange(self.n_features_in_),
        self.categorical_features_,
    )

    if self.categorical_features_.size == self.n_features_in_:
        error_msg = (
            'GeometricSMOTE is not designed to work only with categorical '
            'features. It requires some numerical features.'
        )
        raise ValueError(error_msg)

    return self

_validate_estimators(X)

Validate nearest neighbors estimators.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def _validate_estimators(self: Self, X: NDArray) -> Self:
    """Validate nearest neighbors estimators."""

    # Check random state
    self.random_state_ = check_random_state(self.random_state)

    # Validate strategy
    if self.selection_strategy not in SELECTION_STRATEGIES:
        error_msg = (
            'Unknown selection_strategy for Geometric SMOTE algorithm. '
            f'Choices are {SELECTION_STRATEGIES}. Got {self.selection_strategy} instead.'
        )
        raise ValueError(error_msg)

    # Number of jobs
    n_jobs = None
    if isinstance(self.k_neighbors, NearestNeighbors):
        n_jobs = self.k_neighbors.n_jobs

    # Create nearest neighbors object for positive class
    if self.selection_strategy in ('minority', 'combined'):
        self.nns_pos_ = check_neighbors_object(
            'nns_positive',
            self.k_neighbors,
            additional_neighbor=1,
        )
        self.nns_pos_.set_params(n_jobs=n_jobs)

    # Create nearest neighbors object for negative class
    if self.selection_strategy in ('majority', 'combined'):
        self.nn_neg_ = check_neighbors_object('nn_negative', nn_object=1)
        self.nn_neg_.set_params(n_jobs=n_jobs)

    # Create one hot encoder object
    if self.categorical_features is not None:
        self.ohe_ = OneHotEncoder(
            sparse_output=True,
            handle_unknown='ignore',
            dtype=X.dtype if X.dtype.name != 'object' else np.float64,
        )
        self.ohe_.fit(X[:, self.categorical_features_])

    return self

make_geometric_sample(center, surface_point, truncation_factor, deformation_factor, random_state)

A support function that returns an artificial point.

Parameters:

Name	Type	Description	Default
center	NDArray	The center point.	required
surface_point	NDArray	The point on the surface of the hypersphere.	required
truncation_factor	float	The truncation factor of the algorithm.	required
deformation_factor	float	The defirmation factor of the algorithm.	required
random_state	RandomState	The random state of the process.	required

Returns:

Name	Type	Description
geometric_sample	NDArray	The generated geometric sample.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def make_geometric_sample(
    center: NDArray,
    surface_point: NDArray,
    truncation_factor: float,
    deformation_factor: float,
    random_state: np.random.RandomState,
) -> NDArray:
    """A support function that returns an artificial point.

    Args:
        center:
            The center point.

        surface_point:
            The point on the surface of the hypersphere.

        truncation_factor:
            The truncation factor of the algorithm.

        deformation_factor:
            The defirmation factor of the algorithm.

        random_state:
            The random state of the process.

    Returns:
        geometric_sample:
            The generated geometric sample.
    """

    # Zero radius case
    if np.array_equal(center, surface_point):
        return center

    # Generate a point on the surface of a unit hyper-sphere
    radius = norm(center - surface_point)
    normal_samples = random_state.normal(size=center.size)
    point_on_unit_sphere = normal_samples / norm(normal_samples)
    point: NDArray = (random_state.uniform(size=1) ** (1 / center.size)) * point_on_unit_sphere

    # Parallel unit vector
    parallel_unit_vector = (surface_point - center) / norm(surface_point - center)

    # Truncation
    close_to_opposite_boundary = truncation_factor > 0 and np.dot(point, parallel_unit_vector) < truncation_factor - 1
    close_to_boundary = truncation_factor < 0 and np.dot(point, parallel_unit_vector) > truncation_factor + 1
    if close_to_opposite_boundary or close_to_boundary:
        point -= 2 * np.dot(point, parallel_unit_vector) * parallel_unit_vector

    # Deformation
    parallel_point_position = np.dot(point, parallel_unit_vector) * parallel_unit_vector
    perpendicular_point_position = point - parallel_point_position
    point = parallel_point_position + (1 - deformation_factor) * perpendicular_point_position

    # Translation
    point = center + radius * point

    return point

populate_categorical_features(X_new, neighbors, categories_size, random_state)

A support function that populates categorical features.

Source code in src/imblearn_extra/gsmote/geometric_smote.py

def populate_categorical_features(
    X_new: NDArray,
    neighbors: NDArray,
    categories_size: list[int] | None,
    random_state: np.random.RandomState,
) -> NDArray:
    """A support function that populates categorical features."""
    if categories_size is not None:
        for start_idx, end_idx in zip(
            np.cumsum(categories_size)[:-1],
            np.cumsum(categories_size)[1:],
            strict=False,
        ):
            col_maxs = neighbors[:, start_idx:end_idx].sum(axis=0)
            is_max = np.isclose(col_maxs, col_maxs.max(axis=0))
            max_idxs = random_state.permutation(np.argwhere(is_max))
            col_sels = max_idxs[0]
            ys = start_idx + col_sels
            X_new[start_idx:end_idx] = 0
            X_new[ys] = 1
    return X_new