sppam 0.0.5

A classifier that endeavors to solve the saddle point problem for AUC maximization.

SPPAM - Saddle point problem for AUC maximization

An AUC optimizing binomial classifier.

Contact

Rolf Carlson [e-mail]

Install

Use pip to install sppam.

pip install sppam

Introduction

This is a python implementation of a classifier that endeavors to solve the saddle point problem for AUC maximization. [1]

SPPAM provides classification and prediction for two classes, the binomial case. Small to medium problems are supported. This is research code and a work in progress.

SPPAM is designed for use with scikit-learn pipelines and composite estimators.

Example

from sppam import SPPAM
from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split

Make a classification problem

seed = 42
X, y = make_classification(
    n_samples=30,
    n_features=5,
    n_informative=2,
    n_redundant=2,
    n_classes=2,
    random_state=seed
)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=seed)

Train the classifier

cls = SPPAM().fit(X_train, y_train)

Get the score on unseen data

cls.score(X_test, y_test)

1.0

References

[1] Natole Jr, Michael & Ying, Yiming & Lyu, Siwei. (2019). Stochastic AUC Optimization Algorithms With Linear Convergence. Frontiers in Applied Mathematics and Statistics. 5. 10.3389/fams.2019.00030.