pandas-vet plugin

Project date:

badge PyPI Build Status PyPI - Status PyPI - License

pandas-vet is a plugin for flake8 that provides opinionated linting for pandas code.

It began as a project during the PyCascades 2019 sprints.


Starting with pandas can be daunting. The usual internet help sites are littered with different ways to do the same thing and some features that the pandas docs themselves discourage live on in the API. pandas-vet is (hopefully) a way to help make pandas a little more friendly for newcomers by taking some opinionated stances about pandas best practices. It is designed to help users reduce the pandas universe.

The idea to create a linter was sparked by Ania Kapuścińska’s talk at PyCascades 2019, “Lint your code responsibly!”.

Many of the opinions stem from Ted Petrou’s excellent Minimally Sufficient Pandas. Other ideas are drawn from pandas docs or elsewhere. The Pandas in Black and White flashcards have a lot of the same opinions too.


pandas-vet is a plugin for flake8. If you don’t have flake8 already, it will install automatically when you install pandas-vet.

The plugin is on PyPI and can be installed with:

pip install pandas-vet

pandas-vet is tested under Python 3.5 and 3.6 and should work with later versions as well.


Once installed successfully in an environment that also has flake8 installed, pandas-vet should run whenever flake8 is run.

$ flake8 ...

See the flake8 docs for more information.

For a full list of implemented warnings, see the list below.


pandas-vet is still in the very early stages. Contributions are welcome from the community on code, tests, docs, and just about anything else.

Code of Conduct

Because this project started during the PyCascades 2019 sprints, we adopt the PyCascades minimal expectation that we “Be excellent to each another”. Beyond that, we follow the Python Software Foundation’s Community Code of Conduct.

Steps to contributing

  1. Please submit an issue (or draft PR) first describing the types of changes you’d like to implement.

  2. Fork the repo and create a new branch for your enhancement/fix.

  3. Write code, docs, etc.

  4. We use pytest and flake8 to validate our codebase. The TravisCI integration will complain on pull requests if there are any failing tests or lint violations. To check these locally, run the following commands:

pytest tests
flake8 pandas_vet tests --exclude tests/data
  1. Push to your forked repo.

  2. Submit pull request to the parent repo from your branch. Be sure to write a clear message and reference the Issue # that relates to your pull request.

  3. Feel good about giving back to open source projects.

How to add a check to the linter

  1. Write tests. At a minimum, you should have test cases where the linter should catch “bad” pandas and test cases where the linter should allow “good” pandas.

  2. Write your check function in /pandas-vet/

  3. Run flake8 and pytest on the linter itself (see Steps to contributing)


PyCascades 2019 sprints team

List of implemented warnings (as of v.0.2.0)

PD001: pandas should always be imported as ‘import pandas as pd’

PD002: ‘inplace = True’ should be avoided; it has inconsistent behavior

PD003: ‘.isna’ is preferred to ‘.isnull’; functionality is equivalent

PD004: ‘.notna’ is preferred to ‘.notnull’; functionality is equivalent

PD005: Use arithmetic operator instead of method

PD006: Use comparison operator instead of method

PD007: ‘.ix’ is deprecated; use more explicit ‘.loc’ or ‘.iloc’

PD008: Use ‘.loc’ instead of ‘.at’. If speed is important, use numpy.

PD009: Use ‘.iloc’ instead of ‘.iat’. If speed is important, use numpy.

PD010 ‘.pivot_table’ is preferred to ‘.pivot’ or ‘.unstack’; provides same functionality

PD011 Use ‘.array’ or ‘.to_array()’ instead of ‘.values’; ‘values’ is ambiguous

PDO12 ‘.read_csv’ is preferred to ‘.read_table’; provides same functionality

PD013 ‘.melt’ is preferred to ‘.stack’; provides same functionality

PD015 Use ‘.merge’ method instead of ‘pd.merge’ function. They have equivalent functionality.