SQL Server 2017 introduces the new PREDICT function. This function makes prediction computations much simpler than those that are calculated using R or Python languages, which we looked at in the preceding section. However, the PREDICT function doesn't work with every model that is trained in the arbitrary R (or Python) library.
When the SQL Server started providing machine learning services, new libraries called RevoScaleR for R and RevoScalePy for Python were introduced. These libraries contain their own implementation of several predictive algorithms and also offer the ability to process data in parallel.
Using one of these libraries is a prerequisite when we want to use the PREDICT function. We must also fulfill the following prerequisites:
- We should use one of following algorithms:
- rxLinMod
- rxLogit
- rxBTrees
- rxDtree
- rxForest
- rxFastTrees
- rxFastForest...
