Regression

Regression is the category of machine learning that predicts a numerical value rather than assigning a label or category. Instead of answering "which group does this belong to?", it answers "what number do we expect?" — how much demand for this product next month, what price is this property likely to sell for, how many days until this machine needs maintenance, what is this customer's lifetime value. It is the mechanism behind most business forecasting and financial estimation that AI teams deploy.

Regression models are typically reported by their average error, which can make them look more reliable than they are. A model with a low average error can still have wide variance — meaning individual predictions are frequently far off even when the overall average holds. For decisions that depend on individual forecasts (pricing a deal, scheduling maintenance on a specific asset, approving a specific loan), average accuracy is the wrong measure. The error distribution — how often predictions fall within an acceptable range, and how bad the outliers are — determines whether the model is actually useful for the decision it's supposed to support.