Навчання моделі машинного навчання для клінічних даних IoT і сумісності пристроїв

Abstract

Data exploration, wrangling, and interactive analysis and visualization were made in an integrated way. How to plot feature importance in Python calculated by the XGBoost model was considered. Features engineering in a dataset has been improved with Haar Transform. The area under the receiver operating characteristic curve was increased from 0.44 for the base model to 0.82 for Haar Transform Model.