Welcome to Pandantic’s documentation!

Gone are the days of black-box dataframes in otherwise type-safe code!

Pandantic builds off the Pydantic API to enable validation and filtering of the usual dataframe types (i.e., pandas, etc.)

Installation

Install Pandantic using pip:

$ pip install pandantic

Quick Start

Here’s a simple example demonstrating how to validate a pandas DataFrame:

import pandas as pd
from pydantic import BaseModel

from pandantic import Pandantic

# Define your schema using Pydantic
class EmployeeSchema(BaseModel):
    name: str
    salary: int
    department: str

# Create sample DataFrame with mixed valid/invalid data
df = pd.DataFrame({
    "name": ["Alice Smith", "Bob Jones", 123],          # Last row: invalid name
    "salary": [50000, "high", 60000],                  # Second row: invalid salary
    "department": ["Engineering", "Sales", "Marketing"]
})

# Initialize validator
validator = Pandantic(schema=EmployeeSchema)

# Method 1: Skip invalid rows
df_valid = validator.validate(dataframe=df, errors="skip")
print(f"Valid rows: {len(df_valid)} out of {len(df)}")

# Method 2: Raise error on invalid data
try:
    validator.validate(dataframe=df, errors="raise")
except ValueError as e:
    print(f"Validation error: {e}")

Key Features

Validate DataFrame columns against Pydantic models
Two validation modes: skip invalid rows or raise errors
Full compatibility with Pydantic’s type system and validators
Simple, intuitive API following pandas conventions

Contents