We've covered enough of the basics of Spark now to use our RDDs for machine learning. While Spark handles the infrastructure, the actual work of performing machine learning is handled by an apache Spark subproject called MLlib.
Note
An overview of all the capabilities of the MLlib library are at https://spark.apache.org/docs/latest/mllib-guide.html.
MLlib provides a wealth of machine learning algorithms for use on Spark, including those for regression, classification, and clustering covered elsewhere in this book. In this chapter, we'll be using the algorithm MLlib provides for performing collaborative filtering: alternating least squares.
In Chapter 5, Big Data, we discovered how to use gradient descent to identify the parameters that minimize a cost function for a large quantity of data. In this chapter, we've seen how SVD can be used to calculate latent factors within a matrix of data through decomposition...