Dataset

class Dataset(name: str)

Bases: object

Management class for handling dataset and its properties.

This class serves as a central hub for a dataset, orchestrating the loading of raw data, attribute properties, and generalization hierarchies.

Parameters:

name (str) – The name of the dataset. This should correspond to a directory in k_anonymization/datasets/.

Variables:

• all_datasets (list) – A class-level list containing all instantiated Dataset objects.

• name (str) – The name assigned to the dataset instance.

Methods

_repr_html_	Get the HTML representation of the dataset.
describe	Render interactive dataframe showing dataset info.
reload_df	Read the CSV file of the Dataset from disk.
sample	Create a stratified sample of the dataset.

_repr_html_()

Get the HTML representation of the dataset.

Returns:

str – HTML string for Jupyter display.

describe()

Render interactive dataframe showing dataset info.

reload_df()

Read the CSV file of the Dataset from disk.

sample(n=None, frac=None, seed=None, ignore_index=True)

Create a stratified sample of the dataset.

Samples rows based on the distribution of the target attribute.

Parameters:

• n (int, optional) – Number of rows to sample.

• frac (float, optional) – Fraction of data to sample.

• seed (int, optional) – Random state for reproducibility.

• ignore_index (bool, default True) – If True, the resulting index will be labeled 0, 1, …, n - 1.

Returns:

SampleDataset – A new dataset object containing the sampled records.

Attributes

all_datasets
df	Lazy-load the dataset as a ITableDF.
hierarchies	Lazy-load the generalization hierarchies for this dataset.
info	Generate a summary DataFrame of the dataset's attributes.
is_categorical	A list of booleans indicating if a QID attribute is categorical.
path	Get the absolute file path to the dataset directory.
props	Load and cache the dataset properties from a JSON file.
qids	A list of names of the QID attributes.
qids_categorial	A list of names of the cateforical QID attributes.
qids_idx	The column indices of the QID attributes.
qids_idx_categorial	The column indices of the categorical QID attributes.
qids_idx_numerical	The column indices of the numerical QID attributes.
qids_numerical	A list of names of the numerical QID attributes.
target	The sensitive attribute.

all_datasets = []

df

Lazy-load the dataset as a ITableDF.

Returns:

ITableDF – The underlying data stored in the CSV file.

hierarchies

Lazy-load the generalization hierarchies for this dataset.

Returns:

HierarchiesDict – Hierarchy definitions for the QID attributes.

info

Generate a summary DataFrame of the dataset’s attributes.

Provides a breakdown of each column’s status as a QID, numerical, or categorical attribute, along with the count of unique values.

Returns:

pandas.DataFrame – A summary table with metadata labels as the index.

is_categorical

A list of booleans indicating if a QID attribute is categorical.

Returns:

list

path

Get the absolute file path to the dataset directory.

Returns:

str – The directory path where the dataset assets are stored.

props

Load and cache the dataset properties from a JSON file.

The file is expected to be located at path/props.json.

Returns:

dict – A dictionary containing metadata such as QID indices and categorical flags.

qids

A list of names of the QID attributes.

Returns:

list

qids_categorial

A list of names of the cateforical QID attributes.

Returns:

list

qids_idx

The column indices of the QID attributes.

Returns:

list

qids_idx_categorial

The column indices of the categorical QID attributes.

Returns:

list

qids_idx_numerical

The column indices of the numerical QID attributes.

Returns:

list

qids_numerical

A list of names of the numerical QID attributes.

Returns:

list

target

The sensitive attribute.

Returns:

str