How to Run a Linear Regression on Mac & Windows

Linear regression models how one outcome depends on one or more predictors. Here's how to fit one on Mac & Windows and read what it tells you — coefficients, fit, and the all-important diagnostics.

Regression answers “how does this outcome change as these predictors change?” — and gives you an equation to predict with. On Mac or Windows, Stratum builds the model from menus, no formulas to wire.

Build the model

Open your data and choose Linear Regression.
Pick the response (what you're predicting) and one or more predictors.
Read the fitted model — the coefficients, R-squared, and significance.

Building a linear regression model in Stratum — Pick a response and predictors from menus.

Read the result

Coefficients — each predictor's effect: how much the response changes per unit, holding the others fixed.
R-squared / adjusted R-squared — the share of variation the model explains (adjusted R-squared penalizes adding useless predictors).
p-values — whether each predictor's effect is distinguishable from zero.

A linear regression report in Stratum showing coefficients and R-squared — Coefficients, R-squared, and significance in one report.

Don't skip the diagnostics

A high R-squared doesn't make a model trustworthy. Stratum reports the diagnostics that matter: VIF for multicollinearity, leverage and Cook's distance for influential points, and residual plots to check the model's assumptions. We unpack all of them in Reading Regression Diagnostics.

Predict with it. Once you're happy with the model, add a predicted-value column back to your data with one click.

For the full walk-through — best-subsets, confidence and prediction intervals — see Lesson 17.

Download Stratum

Frequently asked questions

Can I run a regression on Mac & Windows without coding?

Yes — Stratum builds the model from menus: pick a response and predictors and it reports coefficients, R-squared, p-values, and full diagnostics, no R or Python.

What does R-squared tell me?

The fraction of the response's variation the model explains, from 0 to 1. Use adjusted R-squared when comparing models, since it penalizes adding predictors that don't help.

What diagnostics should I check after a regression?

Multicollinearity (VIF), influential points (leverage and Cook's distance), and the residual plots for non-linearity or non-constant variance. Stratum reports all of them.

How do I make predictions from the model?

Add a predicted-value column to your dataset — Stratum does it in one click once the model is fitted.