Skip to main content

Extrapolation

Extrapolation is using a fitted regression line to predict outcomes outside the range of the observed data, where the estimated relationship may no longer hold. It is riskier the further the prediction lies from the data.

ByHoang TruongUpdated

See it move

Loading infographic...

A regression of monthly revenue on floor space, fitted on stores from 50 to 300 square metres, gives Revenue = €3,000 + €80 × floor space. A 200 m² store, inside that range, gives €19,000. A hypothetical 1,000 m² superstore gives €83,000, but nothing in the data confirms revenue keeps rising at €80 per square metre that far out — the prediction is an extrapolation.

Where it fits
SubjectData Analysis & StatisticsCoreTopicSimple Linear Regression & OLSCoreTopicRegression Diagnostics & ProblemsCore

Check yourself

PracticeCORE

A courier company fits a regression of average weekly delivery time (minutes) on weekly parcel volume, using data where volume ranged from 100 to 800 parcels. The fitted line is Delivery time = 25 + 0.05 × Volume. A manager uses it to predict delivery time for a week with 5,000 parcels: 25 + 0.05 × 5,000 = 275 minutes. Which statement correctly evaluates this prediction?

Select an answer to check your understanding.
Extrapolation — Edlintics Glossary