Dataset

verifia.context.Dataset

Represents a dataset containing features and a target label. Provides functionality for sampling, and supports filtering based on domain-specific criteria.

X property

Returns:

Type	Description
NDArray	npt.NDArray: NumPy array containing the feature data.

cat_feature_idxs property

Return the list of indices corresponding to categorical features.

Returns:

Type	Description
	List[int]: A list of indices in self._feature_names for categorical features.

cat_feature_names property

Return the list of categorical feature names.

Returns:

Type	Description
	List[str]: A list of categorical feature names.

data property

Returns:

Type	Description
DataFrame	pd.DataFrame: A copy of the internal DataFrame containing the dataset.

n_samples property

Returns:

Name	Type	Description
int	int	The number of samples in the dataset.

num_feature_idxs property

Return the list of indices corresponding to numeric (non-categorical) features.

Returns:

Type	Description
	List[int]: A list of indices in self._feature_names for numeric features.

num_feature_names property

Return the list of numerical feature names in their original order.

Returns:

Type	Description
	List[str]: A list of feature names that are not categorical.

target_data property

Retrieve the target column from the dataset.

Returns:

Type	Description
Series	pd.Series: A Series corresponding to the target variable specified by self._target_name.

y property

Returns:

Type	Description
NDArray	npt.NDArray: NumPy array containing the target (label) data.

__init__(df, target_name, feature_names=None, cat_feature_names=None)

Initializes the Dataset instance by validating the input DataFrame and inferring categorical features if not provided.

Parameters:

Name	Type	Description	Default
df	DataFrame	Input DataFrame containing the data.	required
target_name	str	Name of the target (label) column.	required
feature_names	Optional[Iterable[str]]	Iterable of all feature names. If None, all features names are automatically inferred from columns in the provided dataframe.	None
cat_feature_names	Optional[Iterable[str]]	Iterable of categorical feature names. If None, categorical features are automatically inferred from columns in feature_names with dtype 'object'.	None

Raises:

Type	Description
ValueError	If the DataFrame does not contain all required columns.

__len__()

Returns:

Name	Type	Description
int	int	The number of samples in the dataset.

feature_data(enforced_data_type=False)

Retrieve the features subset from the dataset.

Returns:

Type	Description
DataFrame	pd.DataFrame: A DataFrame containing only the columns specified in self._feature_names.

sample(n_samples=None, prop_samples=None, replace=False, random_state=None)

Creates a new Dataset instance by sampling rows from the current dataset.

Either n_samples or prop_samples must be provided. If both are provided, n_samples takes precedence.

Parameters:

Name	Type	Description	Default
n_samples	Optional[int]	The exact number of samples to extract. Must be positive.	None
prop_samples	Optional[float]	Proportion of the total samples to extract (between 0 and 1).	None
replace	bool	Whether sampling is done with replacement. Defaults to False.	False
random_state	Optional[Union[int, Generator]]	Seed or Generator for reproducible sampling.	None

Returns:

Name	Type	Description
Dataset	Dataset	A new Dataset instance containing the sampled data.

Raises:

Type	Description
ValueError	If neither n_samples nor prop_samples is provided, if provided values are out of range, or if sampling without replacement and n_samples exceeds available data.

split(primary_split_size, random_state=None)

Split the dataset into a primary split and a secondary split.

Parameters:

Name	Type	Description	Default
primary_split_size	float	The fraction of the dataset to include in the primary split.	required
random_state	Optional[int]	Seed for reproducibility. Defaults to None.	None

Returns:

Type	Description
Tuple[Dataset, Dataset]	Tuple[Dataset, Dataset]: A tuple containing: - A Dataset for the primary split. - A Dataset for the secondary split.