Summary

In this chapter, we have covered the extensible functionalities of the RevoScaleR package to deliver fast and good predictions based on the explored datasets. In the previous chapter Statistical learning with RevoScaleR package, we have covered data exploration, preparation and simple and bi-variate statistics. This chapter showed how RevoScaleR package was designed to work with large datasets (that overcome the limitations of RAM and single CPU), enabling spill to disk and multi threading. The same procedures can be used as well in database instances of R, for delivering the predictions to your business and data residing in the database. We have covered this aspect as well, exploring different algorithms and comparing the solutions. Once you have your model selected, you may want to use the PREDICT clause. which is a new feature in SQL Server 2017 with a slightly altered...