/pyquantifier

A Python package to estimate class prevalence in unlabeled datasets by specifying stability assumptions

Primary LanguageJupyter NotebookMIT LicenseMIT

Python package for ICWSM Tutorial "Prevalence Estimation in Social Media Using Black Box Classifiers"

Setup

Go to the pyquantifier directory,

cd pyquantifier

Setup Python venv and install requirements

python3 -m venv .venv
source .venv/bin/activate
python3 -m pip install --upgrade pip
python3 -m pip install -r requirements.txt

Build and install Python package

python3 -m build
python3 -m pip install --force-reinstall dist/pyquantifier-0.0.1-py3-none-any.whl

Tutorial Jupyter Notebooks

The tutorial includes four sections:

  1. Generating representations of joint distributions

Toxic comments on social media

Estimating the Fraction of Toxic Comments on News Articles

Tutorial Slides

The tutorial slides are available here.

The Calibration-Extrapolation Framework

The Calibration-Extrapolation Framework

References and reading list