Three very different looking dragons stand side by side. On the left, a very small blue dragon with spiky hair and no pattern. In the center, a green striped dragon. On the right, a large purple dragon with spots.

The green striped dragon gives an example calculation using the resulting model from multiple linear regression. The model is shown at the top: "weight = 2.4 + 0.6(spotted) + 0.3*height." The dragon is showing a calculation for a striped dragon that is 5.1 feet tall, with a large name tag saying "Hi, I'm...a 5.1 ft tall striped dragon and my predicted weight is 3.9 tons." An asterisk reads "*dragons are dense."

Two blue dragons stand on scales next to a vertical yardstick showing one slightly taller than the other. Regression estimates are shown at the top as an equation: "weight (tons) = 2.4 + 0.3*height", with explanatory text reading "If all other variables are constant, we expect a 1 foot taller dragon to weight 0.3 tons more, on average."

Two purple dragons stand next to teach other. They look similar, but one has stripes and one has spots. Multiple regression estimates are shown at the top, in the form of an equation: "weight (tons) = 2.4 + 0.6*spotted +..." Text reads. "pattern reference level: striped" and highlighting the 0.6 coefficient text reads "if other variables constant, spotted dragons are expected to weight 0.6 tons more than striped dragons, on average."

A green striped dragon stands on a scale, noting how their weight is actually *greater* than what a model predicts it would be. Text reads "Hi, I'm a 5.1 foot tall striped dragon and my weight is 3.9 tons...but my actual weight is 4.2 tons!" A stylized equation below shows that the residual is calculated by 4.2 - 3.9 = 0.3 tons.

A hoard of multiple different dragons stand on scales, shouting out their residuals based on the model estimates (e.g. "My residual is 0.2 tons! My residual is -0.4 tons! My residual is 0.9 tons!" A histogram to the right shows the distribution of these residuals, with text "Check: are residuals normally distributed?"

Back to Top