This section takes an existing R code that generates the R model and runs against the SQL Server dataset into a workflow, where the model can be refreshed and evaluated on a regular basis, then used for predictive analysis. The following figure shows a typical predictive modeling workflow in an R script:

To integrate this script in SQL Server, you'll need to organize the workflow into three steps:

1. Prepare the data for training
2. Train and save the model using T-SQL
3. Operationalize the model

In this section, the last two steps will use sp_execute_external_script, which invokes an R process. These steps are using the SQL Server extensibility framework, described later on.