etiq.datasets.pandas package

Submodules

etiq.datasets.pandas.base_pandas_dataset module

class etiq.datasets.pandas.base_pandas_dataset.BasePandasDatasetMixin(x_train: DataFrame | None = None, y_train: Series | None = None, prediction_train: Series | None = None, x_valid: DataFrame | None = None, y_valid: Series | None = None, prediction_valid: Series | None = None, x_test: DataFrame | None = None, y_test: Series | None = None, prediction_test: Series | None = None, feature_names: Sequence[str] | None = None, categorical_features: Sequence[str] | None = None, continuous_features: Sequence[str] | None = None, date_features: Sequence[str] | None = None, id_features: Sequence[str] | None = None, history: List[Tuple[str, str, dict]] | None = None, target_name: str | None = None, prediction_name: str | None = None, name: str | None = None, target_categorical: bool = True, prediction_categorical: bool = True)

Bases: object

Implementation of the Simple dataset class using pandas and numpy

property all_feature_names: Tuple[str, ...]

Return the names of the features stored in the dataset

Returns:: A list containing the names of all features in the dataset
Return type:: Tuple[str, …]

property all_features: DataFrame

all_identical(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → bool

property categorical_features: Tuple[str, ...]

Return the names of the categorical features stored in the dataset

Returns:: A list containing the names of all categorical features in the dataset
Return type:: Tuple[str, …]

check_column_ordering_violations(gt_feature: str, lt_feature: str, ids: Tuple[str, ...] | None = None) → Iterable

check_duplicate_row_violations(duplicate_features_subset: Tuple[str, ...] | None = None, ids: Tuple[str, ...] | None = None) → Iterable

check_missing_ids(id_feature: str, primary_ids: List[str] | None = None) → Iterable

property checksum: Dict[str, str]

property continuous_features: Tuple[str, ...]

Returns the names of the continuous features stored in the dataset

Returns:: A list containing the names of all continuous features in the dataset
Return type:: Tuple[str, …]

property data_backend: DataBackend

property date_feature_names: Tuple[str, ...]

Return the names of the features that store dates

Returns:: A list containing the names of the date features
Return type:: Tuple[str, …]

property date_features: DataFrame

property date_features_testing: DataFrame

property date_features_training: DataFrame

property date_features_validation: DataFrame

feature(name: str, mask: str = '') → Series

Returns all data in the dataset for the specified feature.

Parameters:: name (str) – The name of the feature.
Returns:: The data for the specified feature
Return type:: pd.Series

feature_combined_std(other_dataset: AbstractDataset[DataFrame, Series], name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → float

feature_max(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → Any

feature_min(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → Any

property feature_names: Tuple[str, ...]

Return the names of the features stored in the dataset excluding the id and date features

Returns:: A list containing the names of all features in the dataset excluding the id and date features
Return type:: Tuple[str, …]

feature_testing(name: str, mask: str = '') → Series

Returns testing data for the specified feature.

Parameters:: name (str) – The name of the feature.
Returns:: The testing data for the specified feature
Return type:: pd.Series

feature_training(name: str, mask: str = '') → Series

Returns training data for the specified feature.

Parameters:: name (str) – The name of the feature
Returns:: The training data for the specified feature
Return type:: pd.Series

feature_validation(name: str, mask: str = '') → Series

Returns validation data for the specified feature.

Parameters:: name (str) – The name of the feature.
Returns:: The validation data for the specified feature
Return type:: pd.Series

property features: DataFrame

property features_testing: DataFrame

property features_training: DataFrame

property features_validation: DataFrame

filter_count(condition: str, dataset_segment: DatasetSegment = DatasetSegment.ALL) → int

generate_data_profile(group_by: str | None = None) → DataProfile

Create a data profile for the dataset

Parameters:: group_by (str) – Optional field to group the profile by.

get_dataset_checksum()

get_dataset_history()

get_eval(segment: DatasetSegment, condition: str)

get_feature_categories(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → Tuple[Any, ...]

get_feature_profile(afeature: str, feature_vals: List[Any], atype: FeatureType, condition: str | None = None)

get_metric_calculation_params(mask: str = '') → Tuple

get_restricted_features() → Tuple[str, ...]

get_segment_data(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '')

property has_predictions: bool

histogram(feature_name: str, bins: int = 15)

property history: List[Tuple[str, str, dict]]

property id: str

property id_feature_names: Tuple[str, ...]

Return the names of the features that store ids

Returns:: A list containing the names of the date features
Return type:: Tuple[str, …]

property id_features: DataFrame

property id_features_testing: DataFrame

property id_features_training: DataFrame

property id_features_validation: DataFrame

is_categorical_feature(afeature: str) → bool

is_continuous_feature(afeature: str) → bool

is_date_feature(afeature: str) → bool

is_id_feature(afeature: str) → bool

is_numerical(afeature: str) → bool

is_target_categorical() → bool

is_target_continuous() → bool

mean(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → Number

median(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → Number

property name: str

property number_of_date_features: int

property number_of_features: int

property number_of_id_features: int

property number_of_samples: int

property number_of_testing_samples: int

property number_of_training_samples: int

property number_of_validation_samples: int

property overall_checksum: str

property prediction: Series

property prediction_categorical: bool

property prediction_name: str | None

property prediction_testing: Series

property prediction_training: Series

property prediction_validation: Series

quantiles(name: str, quantile_vals: List[float], dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '') → List[Number]

property target: Series

property target_categorical: bool

property target_name: str

property target_testing: Series

property target_training: Series

property target_validation: Series

to_dataframe() → DataFrame: Return this dataset as a Pandas Dataframe instance

etiq.datasets.pandas.base_pandas_dataset.query_eval(df: DataFrame, aquery: str) → ndarray | None

etiq.datasets.pandas.bias_pandas_dataset module

Bases: BasePandasDatasetMixin, BiasDataset[DataFrame, Series]

cache_predictions(predictions: DataFrame, prediction_name: str, condition: str = '') → None

check_column_ordering_violations(gt_feature: str, lt_feature: str, ids: Tuple[str, ...] | None = None) → Iterator

compare_feature(other_dataset: AbstractDataset, afeature: str)

filter(condition: str, dataset_segment: DatasetSegment = DatasetSegment.ALL) → BiasPandasDataset

generate_data_profile(group_by: str | None = None) → DataProfile

Create a data profile for the dataset

Parameters:: group_by (str) – Optional field to group the profile by.

get_dataset_checksum()

get_dataset_history()

get_predictions_cache() → BiasPandasPredictionCache | None

get_protected_metric_calculation_params(mask: str = '') → Any

get_segment_data(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '')

property protected: Series

property protected_name: str

property protected_testing: Series

property protected_training: Series

property protected_validation: Series

to_dataframe(): Return this dataset as a Pandas Dataframe instance

class etiq.datasets.pandas.bias_pandas_dataset.BiasPandasPredictionCache(df: DataFrame, target_label: str, protected_label: str, predicted_label: str)

Bases: PredictionCache[DataFrame, Series]

get_cache_query(condition)

get_data(condition: str = '') → DataFrame

get_predicted(condition: str = '') → Series | str

get_protected(condition: str = '') → Series | str | None

get_target(condition: str = '') → Series | str

set_cache(df: DataFrame, target_label: str, protected_label: str, predicted_label: str)

set_eval_cache(eval_cache: Dict[str, Any])

etiq.datasets.pandas.simple_pandas_dataset module

class etiq.datasets.pandas.simple_pandas_dataset.SimplePandasDataset(x_train: DataFrame | None = None, y_train: Series | None = None, prediction_train: Series | None = None, x_valid: DataFrame | None = None, y_valid: Series | None = None, prediction_valid: Series | None = None, x_test: DataFrame | None = None, y_test: Series | None = None, prediction_test: Series | None = None, feature_names: Sequence[str] | None = None, categorical_features: Sequence[str] | None = None, continuous_features: Sequence[str] | None = None, date_features: Sequence[str] | None = None, id_features: Sequence[str] | None = None, history: List | None = None, target_name: str | None = None, prediction_name: str | None = None, name: str | None = None, target_categorical: bool = True, prediction_categorical: bool = True, is_algorithmic: bool = False)

Bases: BasePandasDatasetMixin, SimpleDataset[DataFrame, Series]

Implementation of the Simple dataset class using pandas and numpy

cache_predictions(predictions: DataFrame, prediction_name: str, condition: str) → None

compare_feature(other_dataset: AbstractDataset[DataFrame, Series], afeature: str)

filter(condition: str, dataset_segment: DatasetSegment = DatasetSegment.ALL) → SimplePandasDataset

get_predictions_cache() → SimplePandasPredictionCache | None

class etiq.datasets.pandas.simple_pandas_dataset.SimplePandasPredictionCache(df: DataFrame, target_label: str, predicted_label: str)

Bases: PredictionCache[DataFrame, Series]

get_cache_query(condition)

get_data(condition: str = '') → DataFrame

get_predicted(condition: str = '') → Series | str

get_protected(condition: str = '') → Series | str | None

get_target(condition: str = '') → Series | str

set_cache(df: DataFrame, target_label: str, predicted_label: str)

set_eval_cache(eval_cache: Dict[str, Any])

Module contents

Bases: BasePandasDatasetMixin, BiasDataset[DataFrame, Series]

cache_predictions(predictions: DataFrame, prediction_name: str, condition: str = '') → None

check_column_ordering_violations(gt_feature: str, lt_feature: str, ids: Tuple[str, ...] | None = None) → Iterator

compare_feature(other_dataset: AbstractDataset, afeature: str)

filter(condition: str, dataset_segment: DatasetSegment = DatasetSegment.ALL) → BiasPandasDataset

generate_data_profile(group_by: str | None = None) → DataProfile

Create a data profile for the dataset

Parameters:: group_by (str) – Optional field to group the profile by.

get_dataset_checksum()

get_dataset_history()

get_predictions_cache() → BiasPandasPredictionCache | None

get_protected_metric_calculation_params(mask: str = '') → Any

get_segment_data(name: str, dataset_segment: DatasetSegment = DatasetSegment.ALL, mask: str = '')

property protected: Series

property protected_name: str

property protected_testing: Series

property protected_training: Series

property protected_validation: Series

to_dataframe(): Return this dataset as a Pandas Dataframe instance

class etiq.datasets.pandas.SimplePandasDataset(x_train: DataFrame | None = None, y_train: Series | None = None, prediction_train: Series | None = None, x_valid: DataFrame | None = None, y_valid: Series | None = None, prediction_valid: Series | None = None, x_test: DataFrame | None = None, y_test: Series | None = None, prediction_test: Series | None = None, feature_names: Sequence[str] | None = None, categorical_features: Sequence[str] | None = None, continuous_features: Sequence[str] | None = None, date_features: Sequence[str] | None = None, id_features: Sequence[str] | None = None, history: List | None = None, target_name: str | None = None, prediction_name: str | None = None, name: str | None = None, target_categorical: bool = True, prediction_categorical: bool = True, is_algorithmic: bool = False)

Bases: BasePandasDatasetMixin, SimpleDataset[DataFrame, Series]

Implementation of the Simple dataset class using pandas and numpy

cache_predictions(predictions: DataFrame, prediction_name: str, condition: str) → None

compare_feature(other_dataset: AbstractDataset[DataFrame, Series], afeature: str)

filter(condition: str, dataset_segment: DatasetSegment = DatasetSegment.ALL) → SimplePandasDataset

get_predictions_cache() → SimplePandasPredictionCache | None