ccao-data / assesspy Goto Github PK

This issue is a clone of ccao-data/assessr#3. It is intended to maintain parity between the two packages.

Recent literature from the IAAO and others has suggested using a Gini-like index for measuring vertical equity in property assessments. We should add such a measurement to this package and the [ccao](ccao-data/ccao] package, then add it to our existing batch of output statistics.

See https://researchexchange.iaao.org/jptaa/vol17/iss2/2/ and https://www.sciencedirect.com/science/article/abs/pii/S1051137723000372.

Python package checklist

• Add new Gini function with assesspy/formulas.py
• Make sure your code updates meet styling and linting requirements (black + pyright)
• Use reStructuredText Markup for documentation and build documentation updates by running sphinx locally.
• Add unit tests, copying the existing tests for cod()
• Run unit tests using tox
• Make pull request
• Cut new release version in GitHub

See ccao-data/data-architecture#581.

While the helper functions for detect_chasing link to detect_chasing's documentation, they aren't available to assesspy users, and their presence on the reference page may be misleading.

The license file here isn't the same as the primary LICENSE of the repo.

For cleaned up version on GitHub bump to 1.0.2

• In package config/setup.py
• In docs
• On pypy
• Create GitHub release + git tag

We currently use setup.py, requirements.txt, and a few other files to setup the package infrastructure for assesspy. Recent Python packages have moved toward consolidating their metadata, dependencies, etc in a pyproject.toml file. Here is an example package: https://github.com/r5py/r5py. We should consider moving to this format in order to simplify package building and management.

The downside of this is likely reduced compatibility with prior version of Python. The upside is reduced complexity + the ability to define build-only dependencies. If we decide to implement this issue, we should make numpy the sole dependencies of this package.

The dark theme for the example vignette generates tables that are not readable by default. See here.

This is with Darkreader off.