Linear Regression is like a good digestive system

A good regression model is like a good digestive system.

A good digestive system gobbles up data and extracts everything that's valuable: proteins, vitamins, carbs...
Whatever comes out must have no nutritional value. 💩

A good regression model gobbles up data and extracts all the information in it.
Whatever comes out must be random, meaningless and worthless.
(Indeed, that's we call it RESIDUAL)

👉 In other words, when you look at your residuals' scatterplot, you should just see a random scaterring of points, with no distinguishable pattern. ♾️

😱 If you CAN spot a pattern (whichever pattern it may be), some information wasn't well digested by your model and ended up where it shouldn't be.

Your model may need to see a doctor.