Multiple Regression Summary Calculator

Paste data (CSV/TSV): first column = y, remaining columns = x1..xk. Empty or non-numeric rows are skipped.

Data includes header row

Include intercept

Confidence level (%)

Predict for a new row (exclude y): Enter predictor values separated by comma/tab/semicolon in the same order as your dataset (excluding the dependent variable).

How to Use

Arrange your data so that the first column is the dependent variable y and the remaining columns are predictors x1, x2, ....
Paste the data in the box above. You may keep or remove a header row and toggle the intercept.
Click Fit Model to compute estimates, standard errors, tests, intervals, VIF, and diagnostics.
Optionally enter a new row of predictor values to get a confidence interval for the mean and a prediction interval.

Tip: Keep predictors on comparable scales to improve numerical stability. Drop or transform perfectly collinear variables if a singular design warning appears.

FAQs

P‑values use the Student’s t distribution for individual coefficients and the F distribution for the overall model. They rely on approximate normality of residuals and correct model specification.

The Variance Inflation Factor quantifies how much the variance of a coefficient is inflated by multicollinearity. Values above ~5–10 suggest the predictor is strongly correlated with others.

This occurs when one or more predictors are exact linear combinations of others (perfect multicollinearity). Remove redundant variables or combine them to resolve the issue.

A confidence interval (CI) covers the mean response at given predictors. A prediction interval (PI) covers a single new observation, so it’s wider because it includes residual variability.

This tool assumes the first column is y. If needed, reorder columns in your data editor so y is first and predictors follow.

They are valid under standard OLS assumptions. For small samples or violations (heteroscedasticity, non‑normality), consider robust or bootstrapped methods outside of this tool.

This version reports classical OLS standard errors. For heteroscedasticity‑robust or clustered errors, export data to a statistical package that supports those estimators.

FAQs

How are p-values computed?

What does VIF measure?

Why do I see a “singular matrix” error?

What’s the difference between CI and PI?

Can I change which column is y?

How reliable are the p‑values and intervals?

Do you provide robust standard errors?

Related Calculators