Data Validation in Python

Pydantic validates and parses data using ordinary type hints, and since version 2 its core is written in Rust. It is the validation layer behind FastAPI and many other libraries, so it is the default choice for application data such as API payloads and settings. jsonschema validates data against JSON Schema documents, which is the better fit when the contract is shared with services written in other languages.

For tabular data, Pandera defines DataFrame schemas in code, with checks on column types, value ranges and custom rules, and works with pandas, Polars and other DataFrame libraries. Great Expectations is a heavier data quality framework: suites of expectations, generated documentation and validation checkpoints inside pipelines, suited to teams that monitor data quality across many sources. Pandera is quicker to adopt inside a single codebase.