API Reference / preprocessing / sampling /

oversampling.py

Random oversampling methods for imbalanced learning.

Simple oversampling techniques that duplicate minority samples.

Classes

RandomOverSampler

class preprocessing.sampling.oversampling.RandomOverSampler(Transformer)

Random over-sampling by duplicating minority samples.

Simply duplicates random minority class samples to balance classes.
Constructor 
__init__(
    self,
    sampling_strategy: Union[float, str, dict] = 'auto',
    random_state: Optional[int] = None,
    shrinkage: Optional[float] = None,
)

Parameters

sampling_strategy
float or str or dict = 'auto'

Sampling strategy:

  • •'auto': balance all classes to match majority
  • •'minority': only oversample minority class
  • •dict: {class_label: target_count}
random_state
int
Random seed.
shrinkage
float
If not None, adds Gaussian noise with this shrinkage factor.
python 
>>> from tuiml.preprocessing.sampling import RandomOverSampler
>>> ros = RandomOverSampler(sampling_strategy='auto')
>>> X_res, y_res = ros.fit_resample(X, y)

Methods

get_parameter_schema (cls) -> Dict[str, Dict]

Return JSON Schema for parameters.

fit (self, X: np.ndarray, y: np.ndarray) -> 'RandomOverSampler'

Fit the sampler.

fit_resample (self, X: np.ndarray, y: np.ndarray) -> Tuple[np.ndarray, np.ndarray]

Fit and resample.

transform (self, X: np.ndarray) -> np.ndarray
__repr__ (self) -> str

ClusterOverSampler

class preprocessing.sampling.oversampling.ClusterOverSampler(Transformer)

Cluster-based oversampling.

Clusters each class and oversamples within clusters to preserve data distribution.
Constructor 
__init__(
    self,
    sampling_strategy: Union[float, str, dict] = 'auto',
    n_clusters: int = 5,
    random_state: Optional[int] = None,
)

Parameters

sampling_strategy
str or dict = 'auto'
Sampling strategy.
n_clusters
int = 5
Number of clusters per class.
random_state
int
Random seed.
python 
>>> from tuiml.preprocessing.sampling import ClusterOverSampler
>>> cbos = ClusterOverSampler(n_clusters=3)
>>> X_res, y_res = cbos.fit_resample(X, y)

Methods

get_parameter_schema (cls) -> Dict[str, Dict]

Return JSON Schema for parameters.

fit (self, X: np.ndarray, y: np.ndarray) -> 'ClusterOverSampler'

Fit the sampler.

fit_resample (self, X: np.ndarray, y: np.ndarray) -> Tuple[np.ndarray, np.ndarray]

Fit and resample.

transform (self, X: np.ndarray) -> np.ndarray
__repr__ (self) -> str