H2O is an open source and distributed machine learning platform that allows you to build machine learning models on large datasets. H2O supports both supervised and unsupervised algorithms and is extremely fast, scalable, and easy to implement. H2O's REST API allows us to access all its functionalities from external programs such as R and Python. H2O in Python is designed to be very similar to scikit-learn. At the time of writing this book, the latest version of H2O is H2O v3.
The reason why H2O brought lightning-fast machine learning to enterprises is given by the following explanation:
H2O provides us with distributed random forests, which are a powerful tool used for classification and regression tasks. This generates multiple trees, rather than single trees. In a distributed random forest, we use the average predictions of both the classification and regression models to reach a final result.