ClassImbalance.jl

<p> <a href="https://doi.org/10.5281/zenodo.3233061"> <img src="https://zenodo.org/badge/DOI/10.5281/zenodo.3233061.svg" alt="DOI"> </a> </p> <p> <a href="https://app.bors.tech/repositories/12287"> <img src="https://bors.tech/images/badge_small.svg" alt="Bors enabled"> </a> <a href="https://travis-ci.org/bcbi/ClassImbalance.jl/branches"> <img src="https://travis-ci.org/bcbi/ClassImbalance.jl.svg?branch=master"/> </a> <a href="https://codecov.io/gh/bcbi/ClassImbalance.jl/branch/master"> <img src="https://codecov.io/gh/bcbi/ClassImbalance.jl/branch/master/graph/badge.svg"/> </a> </p>

Description

This is a package that implements some sampling-based methods of correcting for class imbalance in two-category classification problems. Portions of the SMOTE and ROSE algorithm are adaptations of the excellent R packages DMwR and ROSE.

Installation

To install ClassImbalance, open Julia and run the following two lines:

import Pkg
Pkg.add("ClassImbalance")

SMOTE Example

import ClassImbalance;
y = vcat(ones(20), zeros(180)); # 0 = majority, 1 = minority
X = hcat(rand(200, 10), y);
X2, y2 = smote(X, y, k = 5, pct_under = 100, pct_over = 200)