You can say a lot about what linear regression gets you when you fit a line to nonlinear data. It's a weighted average of the derivative. You don't need a linea

You can say a lot about what linear regression gets you when you fit a line to nonlinear data. It's a weighted average of the derivative. You don't need a linearity assumption.