ML internals: Synthetic Minority Oversampling (SMOTE) Technique
Domino Data Lab
MAY 20, 2021
We present the inner workings of the SMOTE algorithm and show a simple “from scratch” implementation of SMOTE. We use an artificially constructed imbalance dataset (based on Iris) to generate synthetic observations via our SMOTE implementation, and discuss modifications that help SMOTE handle categorical attributes.
Let's personalize your content