Describer

Describer(config)

Generate descriptive statistics for datasets.

Parameters

• config (dict): Descriptive statistics configuration
• method (str): Operation name
  • ‘default’: Use default set of statistics methods
• describe (list): List of statistics methods to apply
  • See supported methods table below
  • For percentile, use dictionary format: {'percentile': k}

Examples

from petsard import Describer


# Using default descriptive methods
desc = Describer(method='default')

# Using custom descriptive methods
desc = Describer(
    method='default',
    describe_method=['mean', 'median', 'std', 'percentile'],
    percentile=0.95,
)

# Analysis
desc.create()
desc_result: dict[str, pd.DataFrame] = desc.eval({'data': df})

# Get results
global_stats: pd.DataFrame = desc_result.get('global')      # Global statistics
column_stats: pd.DataFrame = desc_result.get('columnwise')  # Column-wise statistics
pairwise_stats: pd.DataFrame = desc_result.get('pairwise')  # Pairwise statistics

Methods

create()

Initialize descriptor.

Parameters

None

Returns

None

eval()

Perform descriptive statistical analysis.

Parameters

• data (dict): Data to analyze
  • Format: {'data': pd.DataFrame}

Returns

(dict[str, pd.DataFrame]), varies by module:

• ‘global’: Single row dataframe representing overall dataset desciption results
• ‘columnwise’: Column-level desciption results, each row representing desciption results for one column
• ‘pairwise’: Column pair desciption results, each row representing desciption results for a pair of columns

Appendix: Supported Methods

Overview

Descriptive statistics are divided into three levels:

• Global analysis: Calculate overall dataset properties (e.g., row count)
• Column analysis: Calculate statistics for each column (e.g., mean, standard deviation)
• Pairwise analysis: Calculate relationships between columns (e.g., correlation)

Supported Methods