Robustness of Extreme Gradient Boosting: Regression Tree A Riyawan, K A Notodiputro, and B Sartono
Institut Pertanian Bogor (IPB University)
Abstract
The study aims to examine the robustness of XGBoost machine learning. This research was conducted through the simulation method on data containing outliers. RMSE value measures the best of the model compared with the parametric model, namely the linear regression (least square) model. The results show that XGBoost with the proper parameter selection will be more robust than the linear regression (least square) model.
Keywords: statistical learning, scalable, large datasets
