Docs
Python API
Describer API
eval()

eval()

Evaluate data and generate statistical descriptions or comparisons.

Syntax 

def eval(data: dict) -> dict[str, pd.DataFrame]

Parameters

  • data : dict, required
    • Input data dictionary
    • For describe mode: {"data": pd.DataFrame}
    • For compare mode: {"ori": pd.DataFrame, "syn": pd.DataFrame}
    • Additional keys may be supported based on configuration

Returns

  • dict[str, pd.DataFrame]
    • Dictionary containing evaluation results
    • Keys depend on the granularity of analysis:
      • "global": Overall statistics
      • "columnwise": Column-level statistics
      • "pairwise": Pairwise column relationships (correlations, etc.)

Description

The eval() method performs statistical evaluation on the provided data. The behavior depends on the mode:

  • Describe mode: Generates comprehensive statistical summaries for single dataset
  • Compare mode: Compares multiple datasets and calculates differences

The method automatically determines which statistics to calculate based on data types and configuration.

Basic Examples

Describe Mode 

from petsard.evaluator import Describer
import pandas as pd

# Create sample data
df = pd.DataFrame({
    'numeric': [1, 2, 3, 4, 5],
    'category': ['A', 'B', 'A', 'C', 'B']
})

# Initialize and evaluate
describer = Describer(method='describe')
describer.create()
results = describer.eval(data={'data': df})

# Access results
print(results['global'])      # Overall statistics
print(results['columnwise'])  # Per-column statistics

Compare Mode 

from petsard.evaluator import Describer
import pandas as pd

# Create original and synthetic data
ori_df = pd.DataFrame({
    'value': [1, 2, 3, 4, 5],
    'type': ['A', 'B', 'A', 'C', 'B']
})

syn_df = pd.DataFrame({
    'value': [1.1, 2.2, 2.9, 4.1, 5.2],
    'type': ['A', 'B', 'A', 'C', 'B']
})

# Initialize in compare mode
describer = Describer(
    method='describe',
    mode='compare',
    stats_method=['mean', 'std', 'jsdivergence'],
    compare_method='pct_change'
)
describer.create()

# Evaluate comparison
results = describer.eval(data={'ori': ori_df, 'syn': syn_df})

# Access comparison results
print(results['global'])      # Global comparison score
print(results['columnwise'])  # Column-wise comparisons

Notes

  • Data preparation: Ensure data is properly cleaned before evaluation
  • Memory usage: Large datasets may consume significant memory during evaluation
  • Data types: Different statistics are calculated for numerical vs categorical columns
  • Missing values: NaN values are handled automatically based on the statistical method
  • Performance: Pairwise statistics (correlations) can be computationally expensive for wide datasets
  • Mode consistency: The data structure must match the configured mode (describe vs compare)
create()