Multiple Linear Regression
Model one numerical response using several numerical or categorical predictors.
Current dataset
First row: unique column names. Export Excel files as CSV. Data is shared across analysis pages in this browser.
No dataset loaded.
2. Select your model
Categorical predictors use indicator variables with the first alphabetically sorted category as the reference. Numeric category codes should be converted to text labels before upload. Avoid IDs and near-unique categories.
Check the conditions
Check a linear relationship between the predictors and response, independent errors, roughly normal residuals for inference, and similar residual spread. Predictors must not be exact linear combinations of one another. Use the study design to judge whether population inference is appropriate.
Rows missing any selected value are excluded. The model includes an intercept. Coefficients describe associations while holding the other predictors constant; they do not establish causation.