ML internals: Synthetic Minority Oversampling (SMOTE) Technique
Domino Data Lab
MAY 20, 2021
This renders measures like classification accuracy meaningless. note that this variant “performs worse than plain under-sampling based on AUC” when tested on the Adult dataset (Dua & Graff, 2017). The use of multiple measurements in taxonomic problems. Indeed, in the original paper Chawla et al. References. link] Fisher, R.
Let's personalize your content