This repository contains the R code used to generate examples for this LinkedIn Article.
I generated three situations of a binary classification problem, and compared different performance metrics.
The conclusion was that Phi Coefficient is the most robust metric.
License of this repository: GNU Affero General Public License v3 (AGPL-3.0).