Metadata

Metadata()

Captures dataset properties including data types, missing value percentages, and dataset dimensions. Provides utilities for type inference and SDV metadata conversion.

Parameters

None

Examples

from petsard import Metadata


# Create and build metadata
meta = Metadata()
meta.build_metadata(df)

# Access metadata
col_properties = meta.metadata['col']
dataset_properties = meta.metadata['global']

# Convert for SDV
sdv_meta = meta.to_sdv()

Methods

`build_metadata()`

meta.build_metadata(data)

Create metadata from DataFrame and infer data types.

Parameters

data (pd.DataFrame): Input DataFrame

Returns

None. Updates metadata attribute

`set_col_infer_dtype()`

meta.set_col_infer_dtype(col, dtype)

Set the inferred data type for a specific column.

Parameters

col (str): Column name
dtype (str): Inferred data type
- Values: ’numerical’, ‘categorical’, ‘datetime’, ‘object’

Returns

None. Updates column’s infer_dtype

`to_sdv()`

Convert metadata to SDV compatible format.

Parameters

None.

Returns

dict: SDV formatted metadata

Attributes

metadata: Nested dictionary containing dataset properties:
- col: Per-column properties
  - dtype: pandas data type
  - na_percentage: Proportion of NA values
  - infer_dtype: Inferred data type (’numerical’, ‘categorical’, ‘datetime’, or ‘object’)
- global: Dataset-wide properties
  - row_num: Number of rows
  - col_num: Number of columns
  - na_percentage: Overall proportion of NA values

Reporter Describer