Predictive, rather than generative, supervised learning technique that outputs a number, for example rentals or revenue.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains Regression in context, with comparison tables and the common traps.
Terms in this definition
- Supervised learning
Training a model on examples whose correct output is labelled so it learns to predict that output; regression and classification are typical cases.
Related terms
- ML
Building models that infer patterns from data so they can predict outcomes, for instance through regression or classification, using tools like Azure Machine Learning or Synapse Spark pools. Predictive ML is distinct from generative AI.
- Query Store
Keeps a history of query text, plans and runtime statistics, making a plan regression visible and fixable by forcing an earlier plan. It is enabled from the start in Azure SQL and Fabric's SQL database, and for databases newly created on SQL Server 2022 onwards.
- Regressed Queries
Lists queries whose metrics have deteriorated; this SSMS Query Store report includes buttons to force or unforce a plan when fixing a regression.
- RMSE
Root mean squared error, a regression metric: take the differences between predicted and actual values, square them, average them and take the square root. Smaller values are better.
- Synthetic evaluation dataset
A versioned dataset produced by the Foundry data generation service (preview), containing multi-turn simulation seeds or single-turn Q&A pairs derived from a prompt, reference file or agent definition. It serves as a baseline before launch and for regression.