Not all relationships are linear, however. Some data describe relationships that are curved (such as polynomial relationships) while still other data cannot be parameterized. Mathematically ...

The curve estimation procedure can be used to identify the nature of the functional relationships at play in your data, so you can choose the correct regression model, whether linear or nonlinear.

Generalized Linear Models (GLMs) provide an extension to OLR since response variables can be continuous or discrete (e.g. binary or frequency). What are GLMs? When should we use them? How GLM works.