In this chapter, we will use the h2o package to build and train forecasting models with the use of ML models. H2O is an open source, distributed, and Java-based library for machine learning applications. It has APIs for both R (the h2o package) and Python, and includes applications for both supervised and unsupervised learning models. This includes algorithms such as deep learning (DL), gradient boosting machine (GBM), XGBoost, Distributed Random Forest (RF), and the Generalized Linear Model (GLM).
The main advantage of the h2o package is that it is based on distributed processing and, therefore, it can be either used in memory or scaled up with the use of external computing power. Furthermore, the h2o package algorithms provide several methods so that we can train and tune machine learning models, such as the cross-validation method and the built-in grid search function...