Multiple Regression screenshot

Multiple Regression

Ordinary least squares with up to four predictors: coefficients, standard errors, t, p and adjusted R squared.

Privacy First
0 Runs
Updated 2026-11-06
No upload
Runs in this tab
No signup

What this page computes

Put one observation per line as y, x1, x2 and so on, comma-separated. The page solves the normal equations XᵀXβ = Xᵀy by Gauss–Jordan elimination and prints each coefficient with its standard error, t statistic and two-tailed p, plus R², adjusted R², the residual standard error, the degrees of freedom and the row count. Standard errors come from σ²(XᵀX)⁻¹ with σ² = RSS/(n − p). Figures print to six significant figures.

Collinear predictors make the system singular, and the page refuses that rather than printing unstable coefficients; rows must all have the same number of columns, and there must be more rows than coefficients. The page does not select variables, does not standardise the predictors and reports no diagnostics.

The fit runs in this tab, so the same rows always give the same coefficients.

The page fits a least-squares regression of the first column on up to four predictors and prints R², adjusted R², the residual standard error and a per-coefficient block of estimate, standard error, t and two-tailed p. Six rows with two predictors give R² = 0.922078, adjusted R² = 0.87013 and a residual standard error of 0.6742 on 3 degrees of freedom. It takes at most four predictors, and an outcome column that is constant is refused because there is no variation to explain.

Fit a multiple regression from rows of y and up to four predictors, with each coefficient’s two-tailed test.

How to use it

Paste one row per line: the response first, then each predictor, comma-separated. Every row must carry the same number of columns, and a row with more than four predictors is refused rather than truncated. The coefficient block labels the intercept and then b1 upward in column order.

Degrees of freedom and the two R² values

The t statistics use the residual degrees of freedom n − p − 1, where p counts predictors, and each p-value is two-tailed against the hypothesis that the coefficient is zero. Adjusted R² subtracts a penalty for the predictor count, so it falls when a predictor is added that does not earn its place — which is why it is printed beside the plain R² rather than instead of it. Nothing is dropped, standardised or selected automatically.

What to Expect

Best for

  • Fitting a small model to a table you already have

What this tool keeps

  • Nothing leaves this tab: the arithmetic runs in the page you are reading, and a reload discards both the input and the result.

Sample input

5, 1, 0 7, 2, 0 4, 1, 1 6, 2, 1 8, 3, 1 9, 3, 2

Sample output

R² = 0.922078, adjusted R² = 0.87013 n = 6 rows, 2 predictor(s), df = 3 residual standard error = 0.6742 coefficients (estimate, SE, t, two-tailed p): intercept = 2.45455, SE 0.732933, t 3.34894, p 0.0440936 b1 = 2.13636, SE 0.41907, t 5.09787, p 0.0145946 b2 = -0.272727, SE 0.49793, t -0.547723, p 0.622002

Frequently asked questions

How is the table laid out?

One row per line with the response first and the predictors after it, comma-separated. Every line needs the same number of columns, and more than four predictors is refused rather than cut down.

Why print adjusted R² as well?

Because plain R² never falls when a predictor is added, even a useless one. Adjusted R² subtracts a penalty for the predictor count, so the two lines together show whether an extra column earned its place.

Is my table uploaded?

No. The fit and every test are computed in this page from the pasted rows, and nothing is posted to a server or stored.

Developer program

Build custom tools for your store

Join the SDK program to create useful ecommerce utilities for your store.