---
id: data-science-types
version: "0.2.23"
license: Apache License 2.0
license_treatment: permissive
maintenance: abandoned
---

# data-science-types — Type stubs for Python machine learning libraries

License: permissive · Maintenance: abandoned · Downloads: 181.0K/mo

## What it is and what it does

This is a stub-only package that teaches mypy, pytype, and PyCharm how to type-check code using NumPy, pandas, and Matplotlib. It does not add runtime behavior—it only provides type information for static analysis. The stubs are PEP-561 compliant, meaning type checkers will automatically recognize them when the package is installed.

The package is explicitly a work in progress and incomplete. Many functions lack type annotations in the stubs. As a result, you may encounter type checker errors claiming functions do not exist when they actually do. The project was archived and has not been maintained since 2021-02-16, so new library features and API changes are not reflected in the stubs.

Use it for:

- Catch type errors in array operations before runtime, such as incorrect dtype assignments or incompatible arithmetic.
- Enable IDE autocompletion and type hints for data structures in editors like PyCharm.
- Validate plotting code with type checking to catch incorrect argument types to functions.
- Add static type safety to data science scripts without waiting for upstream libraries to ship their own stubs.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Provides PEP-561-compliant type stubs for NumPy, pandas, and Matplotlib, enabling mypy and other type checkers to recognize and validate types in these libraries.

No. The package is abandoned and has not been maintained since 2021-02-16. While it has no runtime dependencies and installs easily, its incomplete stub coverage combined with lack of maintenance means it will produce false negatives and false positives. Modern versions of the supported libraries have either shipped their own stubs or improved type support, making this package obsolete for current projects.

## Install

pip install data-science-types
uv add data-science-types
poetry add data-science-types

## Installing data-science-types

Before you install: Installation is frictionless with no runtime dependencies. However, the package is abandoned and has not been maintained since 2021-02-16, which means type coverage gaps will persist and newer library versions may lack stub support.

License in practice: Licensed under Apache License 2.0 (permissive), so you can use it freely in commercial and open-source projects without restriction.

Quickstart:

pip install data-science-types

# Then type checkers will recognize types in the supported libraries
# Example from the package documentation:
arr1: np.ndarray[np.int64] = np.array([3, 7, 39, -3])  # OK
arr2: np.ndarray[np.int32] = np.array([3, 7, 39, -3])  # Type error

Requires Python 3.6 or later. Type coverage is incomplete; many functions lack stubs, so type checkers may report false 'function does not exist' errors.

Verify before relying:

- Current compatibility with recent versions of NumPy, pandas, and Matplotlib given the last release was 2021-02-16
- Whether stub coverage has improved or regressed relative to upstream library API changes since abandonment
- How much of NumPy and pandas functionality is actually covered by the stubs

## Package facts

- License: Apache License 2.0 (permissive)
- Python support: supports_current
- Install friction: low
- Maintenance: abandoned
- Downloads: 181.0K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

mypy type stubs numpy pandas, type hints for data science libraries, pep 561 type stubs, static type checking data science, matplotlib type annotations, type-stubs, abandoned

[View on SkillFed](https://skillfed.io/packages/data-science-types) · [View on PyPI](https://pypi.org/project/data-science-types/)