Using Platt's scaling for calibration after undersampling -- limitations and how to address them

Open in new window