Free regression and correlation calculator online
This regression calculator answers a question engineers ask every week: does the result depend on this setting, and by how much? Enter pairs of values, such as hardness and furnace temperature, and get the fitted line, the equation, R², the correlation coefficient and a clear verdict.
With two to four X columns it runs a multiple regression with p-values for every factor, the variance inflation factor (VIF) to spot correlated inputs, and a correlation matrix, without Excel add-ins or a statistics licence.
How it works in 4 steps: Regression and Correlation
- CollectY and X pairsone row per part
- FitY = 18 + 0.21 Xleast squares line
- JudgeR² 86 %, p < 0.001strength and significance
- ActConfirm with DOEcorrelation is not proof of cause
How the Regression and Correlation works
The model is fitted by ordinary least squares. Each coefficient gets its standard error, t-value, p-value and confidence interval; the whole model gets R², adjusted R² and the F-test. Standardised residuals above 2 are listed as unusual observations.
All results are checked against SciPy and NumPy in the automated test suite.
When to use it
Use regression to check whether a result follows a process setting (temperature, speed, pressure, material property), to build a simple prediction of a characteristic, to compare a new gauge with a reference, or to back an engineering decision with data before planning a DOE.
PDF report in one click
Export the analysis as a PDF report with your header, the verdict, the equation, the fitted line and residual plots, the coefficient table and the ANOVA table.
Private: everything runs in your browser
The regression is calculated in your browser. Your data is never uploaded, and no account is needed.
How to use the regression calculator
- Enter the data: one row per part or sample, the result (Y) first and the factors (X) after it.
- Choose the number of X columns: one for a simple fitted line, two to four for multiple regression.
- Read the verdict: the equation, R² and whether the relationship is statistically significant.
- Check the residual plot for curves, funnels and unusual points before you trust the model.
Typical uses: hardness against furnace temperature, torque against thread length, a gauge reading against the reference (see also the linearity study), or scrap rate against line speed. When the regression points to a factor, confirm it with a DOE.
Frequently asked questions
›What is the difference between correlation and regression?
Correlation (Pearson r, from −1 to +1) says how strongly two variables move together. Regression gives the equation Y = b0 + b1·X, so you can predict Y and see how much it changes per unit of X.
›What does R² tell me?
The share of the variation in Y that the X columns explain. R² = 80 % means 80 % of the scatter in Y follows X; the other 20 % is noise or factors that are not in the data.
›Does a significant regression prove cause and effect?
No. Two values can move together because a third factor drives both. Use the regression to find candidates and confirm them with a planned experiment (DOE).
›What is VIF?
The variance inflation factor shows how strongly an X column is correlated with the other X columns. Above about 5 the separate effects of those columns cannot be told apart reliably.
›Are the results validated?
Yes. Coefficients, standard errors, p-values, R² and the F-test are checked against SciPy and NumPy least squares in the automated test suite.