Oracle Machine Learning for SQL on Amazon RDS: Build machine learning models entirely in SQL

This article demonstrates how to build and deploy a credit risk scoring model entirely in SQL using Oracle Machine Learning for SQL (OML4SQL) on Amazon RDS for Oracle, with no external ML platforms or data movement required.

• OML4SQL includes AutoML that automatically evaluates multiple algorithms and selects the best performer for your data
• Train classification models using the DBMS_DATA_MINING package with automatic data preparation (PREP_AUTO)
• Score new data using standard SQL PREDICTION() and PREDICTION_PROBABILITY() functions
• Works on both Oracle Database 26ai (CDB architecture) and Oracle Database 19c (non-CDB architecture)
• Set up secure private VPC infrastructure with no public internet access using EC2 and Systems Manager Session Manager
• Model training is CPU-intensive but scoring is lightweight with minimal query overhead

In-database machine learning eliminates data redundancy, reduces latency, and simplifies deployment for classical ML tasks like credit risk scoring and fraud detection on Oracle RDS.