CalculatorToolset
Statistics calculation guides

Fit a linear regression and inspect residuals

Calculate slope and intercept, interpret R² narrowly, and check residual behavior before using a fitted line.

Intent
Use this guide when paired numeric observations need a descriptive straight-line fit rather than a causal claim.
Reviewed
Reading time
9 minutes

What this guide helps you decide

Ordinary least squares chooses the line that minimizes the sum of squared vertical residuals under its stated setup. The slope describes modeled change in y per unit x, while the intercept is the fitted value at x = 0 and may not be meaningful outside the observed range.

R² summarizes the fraction of observed y variation accounted for by the fitted line in the sample. A high R² does not establish causation, validate extrapolation, or prove that residual assumptions hold.

Fit, then diagnose

Calculate means, covariance-style cross-products, slope, and intercept from valid pairs. Generate fitted values and residuals for every point, then inspect nonlinearity, changing spread, influential observations, and extrapolation range.

  1. Pair x and y values without dropping one side independently.
  2. Check for constant x, which makes the slope denominator zero.
  3. Calculate slope, intercept, fitted values, residuals, and R².
  4. Inspect residual patterns and restrict predictions to a justified range.
Worked scenario

Worked scenario: four paired observations

Fit y to x for the pairs (1,2), (2,3), (3,5), and (4,8).

  1. The least-squares slope is 2 and the intercept is −0.5, giving ŷ = −0.5 + 2x.
  2. Fitted values are 1.5, 3.5, 5.5, and 7.5; residuals alternate +0.5, −0.5, −0.5, +0.5.
  3. R² is approximately 0.9524 for these four observations.

Outcome: The line describes the small sample closely, but four points are not evidence of causality or safe prediction beyond x = 1 to 4.

Regression-review checklist

  • Keep x/y pairs aligned and report n.
  • Reject a zero-variance predictor.
  • Inspect residuals, not only R².
  • Label slope with y-units per x-unit.
  • Avoid unsupported extrapolation and causal language.

Limits and responsible use

  • A simple linear fit can be distorted by outliers, nonlinear relationships, dependence, unequal variance, and measurement error.
  • The calculator does not establish causation, select variables, validate study design, or replace domain-specific statistical analysis.

Authoritative references

These links support the definitions, conventions, or safety boundaries used in this guide. CalculatorToolset wrote the explanation and example independently.

Frequently asked questions

Can R² be high for a bad model?

Yes. A restricted range, trend, influential points, or nonlinear pattern can produce a high value while the model remains unsuitable for the intended use.

Why might the intercept be meaningless?

If x = 0 is outside the observed or physically possible range, the fitted intercept is an algebraic component rather than a useful real-world estimate.