Comparing Pandera vs Pydantic

pandera_vs_pydantic.py

import pandas as pd
import time
from pydantic import BaseModel, ValidationError, Field
import pandera as pa
from pandera import Column, DataFrameSchema


# Generate a synthetic DataFrame
def generate_data(n: int) -> pd.DataFrame:
    data = {
        "id": list(range(1, n + 1)),
        "name": [f"name_{i}" for i in range(1, n + 1)],
        "age": [i % 100 for i in range(1, n + 1)],
        "salary": [i * 1000.0 for i in range(1, n + 1)]
    }
    return pd.DataFrame(data)


# Pydantic model for validating a row
class RowModel(BaseModel):
    id: int = Field(..., ge=1)
    name: str
    age: int = Field(..., ge=0, le=100)
    salary: float = Field(..., ge=0.0)


# Pandera schema for validating the entire DataFrame
schema = DataFrameSchema({
    "id": Column(int, checks=pa.Check.ge(1)),
    "name": Column(str),
    "age": Column(int, checks=[pa.Check.ge(0), pa.Check.le(100)]),
    "salary": Column(float, checks=pa.Check.ge(0.0))
})


# Validate rows using Pydantic
def validate_with_pydantic(df: pd.DataFrame) -> bool:
    valid = True
    for row in df.to_dict(orient="records"):
        try:
            RowModel(**row)
        except ValidationError as e:
            valid = False
            print(f"Pydantic Validation Error: {e}")
    return valid


# Validate entire DataFrame using Pandera
def validate_with_pandera(df: pd.DataFrame) -> bool:
    try:
        schema.validate(df)
        return True
    except pa.errors.SchemaError as e:
        print(f"Pandera Validation Error: {e}")
        return False


# Benchmark function
def benchmark(n: int) -> None:
    df = generate_data(n)

    # Benchmark Pydantic validation
    start_time = time.time()
    pydantic_valid = validate_with_pydantic(df)
    pydantic_time = time.time() - start_time

    # Benchmark Pandera validation
    start_time = time.time()
    pandera_valid = validate_with_pandera(df)
    pandera_time = time.time() - start_time

    # Results
    print(f"Pydantic validation passed: {pydantic_valid}, Time taken: {pydantic_time:.6f} seconds")
    print(f"Pandera validation passed: {pandera_valid}, Time taken: {pandera_time:.6f} seconds")


# Run the benchmark with 10000 rows
if __name__ == "__main__":
    benchmark(10000)