What do slope and intercept mean in a linear model?
The intercept is the modeled output at input zero; the slope is the predicted change in output per unit of input. Context decides whether either interpretation is sensible.
The result first, followed by the reasoning and a clean verification.
Updated July 13, 2026Always attach units and ask whether x = 0 lies inside the meaningful domain.
Slope carries two units
If x is measured in hours and y in dollars, the slope is dollars per hour. A positive value predicts growth; a negative value predicts decline.
A model's slope describes its average linear pattern, not necessarily the exact change in every individual observation.
The intercept may be meaningful—or merely algebraic
The y-intercept predicts the output at x = 0. In a cost model it may be a fixed starting fee; in a model of adult height versus age, age zero may be outside the data's useful range.
Do not force a story onto b when the input zero is impossible or far beyond the observed data.
Prediction needs a domain
Interpolation predicts within the observed input range and is usually safer than extrapolation beyond it. Real relationships may curve or change after the data ends.
State units, input range, and practical constraints alongside the equation so the model is not mistaken for a universal law.
A bike rental costs C(h) = 12h + 8 dollars for h hours. Interpret 12 and 8.
Each additional hour adds twelve dollars.
Eight dollars is the modeled starting fee.
Use the model for a five-hour rental.
Common mistakes
- Giving the slope without units.
- Calling the intercept meaningful without checking x = 0.
- Extrapolating far beyond the data without a warning.
Three takeaways
- Slope is a contextual rate.
- The intercept is the output at zero.
- A model needs a sensible domain.