Data Description

Before data synthesizing, you might want to understand your data first. PETsARD provides a description module that analyzes your data at three different granularities.

Click the below button to run this example in Colab:

---
Loader:
  data:
    filepath: 'benchmark/adult-income.csv'
Describer:
  summary:
    method: 'default'
Reporter:
  save_report_global:
    method: 'save_report'
    granularity: 'global'
  save_report_columnwise:
    method: 'save_report'
    granularity: 'columnwise'
  save_report_pairwise:
    method: 'save_report'
    granularity: 'pairwise'
...

Statistics at Three Granularities

Global Statistics (global)

Provides basic information about the entire dataset
Includes total row count, column count, and missing value count

Column Statistics (columnwise)

Individual statistics for each column
Numerical columns: mean, median, standard deviation, min/max values, quartiles, etc.
Categorical columns: number of categories, missing value count

Pairwise Statistics (pairwise)

Analyzes relationships between columns
Primarily provides correlation coefficients between numerical columns

Use Cases Data Preprocessing