Sql Server and R for Real-time Predictions
summary
Shows how to serialize R models into a SQL Server table, version them with temporal tables, and call them back with sp_execute_external_script to return realtime predictions (example: predicted vs actual arrival delay for top five records). Uses Microsoft Machine Learning Server/Revorx C++-accelerated algorithms (eg rxLinModel for scalable linear fits), supports Docker/SQL DB deployment, automatic scheduled retrains, and a sample ms_sql_ml repo with stored procedures, feature engineering and prediction logging that demonstrably returns live predictions.
AI-written summary · about
chapters (14)
- 0:00Start
- 5:19What's new
- 8:19High-level reasons to use it
- 10:26ML integrated with SQL
- 13:40Pushing to remote servers
- 14:25A basic execution
- 16:25Model storage table
- 17:43Model UPSERT stored procedure CREATE PROCEDURE modelUpsert
- 17:55Add some data
- 18:10Produce a model
- 19:14Use model in SQL
- 20:32Produce a native model
- 22:51Other Microsoft ML things
- 24:09Conclusion- Thank you
related talks
37mIn database Machine Learning with Python in SQL Server
53mReal-Time Big Data Analytics with R
28mCharge Your Database with Predictive ML Capabilities using Open-Source
21mTidypredict with recipes, turn workflow to SQL, spark, duckdb and beyond
56mDeveloping AI Applications with SQL AI: Ground to Cloud to Fabric
41mData Engineers: Respect the Resiliency of SQL
30mSimplify Big Data & AI on Spark and Ray with MLSQL
22mPredict the future (or don't) with machine learning in applications