OversampleQA Documentation

OversampleQA is a validation toolkit for oversampling methods in imbalanced classification.

Quick Start

from imblearn.over_sampling import SMOTE
from sklearn.datasets import make_classification

from oversampleqa import validate_oversampling

X, y = make_classification(n_samples=1000, weights=[0.9, 0.1], random_state=42)

error_rate = validate_oversampling(
    X=X,
    y=y,
    minority_label=1,
    oversampler=SMOTE(random_state=42),
)

print(f"Error rate: {error_rate:.3f}")

Installation

pip install oversampleqa

For development:

git clone https://github.com/diogoribeiro7/OversampleQA.git
cd OversampleQA
poetry install

Contents

• Installation
• Quick start
• Concepts
• User guide
• Decision guide for choosing between error rates, calibration, two-sample tests, fidelity, downstream utility and benchmark rankings.
• Limitations
• Production audit workflow
• API overview
• API reference
• Tutorials
• Examples
• FAQ