Machine learning is the science of making machines work without programming predefined rules and learn from data. It is used to build regression, classification, clustering, anomaly detection, and recommender-based systems. It involves training or fitting a model on historical data and using the trained model to make predictions for new data.

Spark provides you with the MLlib library which is a RDD based API and ML pipelines which is a DataFrames based API to build machine learning applications. MLlib is in maintenance mode from version 2.0 and it will be deprecated and discontinued in upcoming releases. ML pipelines will be the mainstream API. Apache Mahout was a machine learning library built on top of Hadoop, which is now integrated with Spark to provide in-memory performance and avoid scalability issues. H2O is an open source project that has powerful machine learning and deep learning algorithms that provide integration with Spark as a Sparkling Water product.

Both Hadoop and Spark platforms provide higher scalability and performance to machine learning algorithms.

In the next chapter, we will build recommendation systems.