COMPEL Glossary / regression
Regression
Regression is a supervised learning task that predicts a continuous numerical value rather than a discrete category.
What this means in practice
Examples include forecasting next quarter's revenue, estimating a property's market value, predicting remaining equipment lifetime, or projecting customer lifetime value. Regression powers demand forecasting, pricing models, financial projections, and predictive maintenance -- all common enterprise AI use cases with clear financial returns. For transformation leaders, regression models are attractive because their accuracy is directly measurable against actual outcomes, making ROI calculation straightforward. However, regression accuracy depends heavily on the availability and quality of historical data with known outcomes, which should be assessed during the COMPEL Calibrate stage.
Why it matters
Regression models power demand forecasting, pricing, financial projections, and predictive maintenance, all common enterprise use cases with clear financial returns. Their accuracy is directly measurable against actual outcomes, making ROI calculation straightforward and business case development compelling. However, regression accuracy depends heavily on historical data quality, which must be assessed before committing to development investment.
How COMPEL uses it
During the Calibrate stage, data quality for regression targets is assessed as part of Data Readiness (Domain 6). Regression use cases are evaluated in the Model stage for feasibility based on available historical data with known outcomes. The Produce stage implements regression models within the MLOps framework, and the Evaluate stage compares predictions against actuals to validate the value thesis.
Related Terms
Other glossary terms mentioned in this entry's definition and context.