OpenEconometrics

openecon is an Apache-2.0 Python library for statistical and econometric analysis. It provides model fitting, persistent JSON results, saved prediction, LaTeX tables and offline charts. The package is an alpha with bounded method and option coverage; full Stata or other vendor parity has not been established.

This is a package-description refresh of SDK 0.3.20a1 and charts 0.3.2a1. Econometric and chart source code is unchanged apart from the package version constants; the alpha method and option scope is unchanged.

Install this package version

python -m pip install 'openecon==0.3.20a2' 'openecon-charts==0.3.2a2'
python -m pip check

Python must be at least 3.11 and below 3.15. The SDK pins charts 0.3.2a2 exactly. Charts also works as a standalone package with no required Python dependencies:

python -m pip install 'openecon-charts==0.3.2a2'

The base SDK installation does not require the web server, CLI, MCP service or additional Excel/Stata file readers. Optional application layers are described in the distribution guide. For example, Excel and Stata readers use python -m pip install 'openecon[files]==0.3.20a2'.

Fit and save a result

import openecon as oe

frame = oe.example()
result = oe.ols(
    data=frame,
    y="wage",
    x=["education", "experience"],
    covariance="HC3",
    device="cpu",
)
print(result.summary())

saved = result.model_dump_json(indent=2)
restored = oe.ResultBundle.model_validate_json(saved)
print(restored.to_latex())

chart = oe.plot.scatter(data=frame, x="education", y="wage")
chart.save_html("wage.html")

Charts embed their scripts, styles and fonts for offline HTML. Optional network PDF/SVG/PNG export uses an installed Chrome, Chromium or Edge; it does not download a browser or open a user's existing profile.

Scope and provenance

SDK 0.3.20a2 and charts 0.3.2a2 form one source-based package pair. The corresponding public source snapshot records the frozen source commit and file inventory in SOURCE-MANIFEST.json. Release provenance identifies the exact wheel/sdist hashes and the scope of their installation checks. Numerical tests, installed-package checks, public release access and registry installation are separate evidence layers.

Installing these Python packages does not rebuild a native desktop installer. Device support follows each method's recorded contract; physical CUDA, installed desktop acceptance, Developer ID signing and notarization have separate evidence.

• Public repository and releases
• Estimator and resource scope
• OLS and inference options
• Public source and fixture boundary
• Apache-2.0 license
• Third-party notices
• Security reporting