abrahamnunes/newton-raphson

An implementation of logistic regression with the Newton-Raphson method.

Newton-Raphson Logistic Regression

A simple and fast Python 3+ implementation of logistic regression for association analyses using the Newton-Raphson method.

• Free software: MIT license
• Documentation: https://newton-raphson.readthedocs.io.

Installation

pip install git+https://github.com/abrahamnunes/newton-raphson

Basic use case

import numpy as np
from sklearn.datasets import make_classification
from newton_raphson import logistic_regression

# Generate synthetic data
X, y = make_classification(n_samples=100, n_features=5)

# Perform logistic regression
res = logistic_regression(X, y)

# Print results
res.summary()

To-Do

• Add unit tests for Hessian scaling factor
• Add unit test for Hessian conditioning
• Add capability to monitor each optimization iteration step

How to Cite

If you use Newton-Raphson in your work, we would very much appreciate the citation, which can be done as follows:

• Abraham Nunes (2018). Newton-Raphson Logistic Regression. Zenodo. http://doi.org/10.5281/zenodo.1211725