Installation

Requirements

• Python 3.10 or higher

• NumPy >= 1.20.0

• SciPy >= 1.5.0

• scikit-learn (for some metrics and utilities)

Install from PyPI

The easiest way to install optimal-classification-cutoffs is using pip:

pip install optimal-classification-cutoffs

Install from Source

For the latest development version, you can install directly from the GitHub repository:

pip install git+https://github.com/finite-sample/optimal-classification-cutoffs.git

Development Installation

For development purposes, clone the repository and install in editable mode:

git clone https://github.com/finite-sample/optimal-classification-cutoffs.git
cd optimal-classification-cutoffs
pip install -e .

To install with example dependencies (for running example notebooks):

pip install -e ".[examples]"

Verification

To verify your installation, run:

import optimal_cutoffs
print(optimal_cutoffs.__version__)

# Quick test
import numpy as np
from optimal_cutoffs import get_optimal_threshold

y_true = np.array([0, 0, 1, 1])
y_prob = np.array([0.1, 0.4, 0.35, 0.8])
threshold = get_optimal_threshold(y_true, y_prob, metric='f1')
print(f"Installation successful! Optimal threshold: {threshold}")

Troubleshooting

Import Errors

If you encounter import errors, ensure all dependencies are installed:

pip install numpy>=1.20.0 scipy>=1.5.0 scikit-learn

Version Conflicts

For clean environment setup, consider using virtual environments:

python -m venv optimal_cutoffs_env
source optimal_cutoffs_env/bin/activate  # On Windows: optimal_cutoffs_env\Scripts\activate
pip install optimal-classification-cutoffs