Nonlinear Modeling: Something Fishy
In this lesson by Douglas Whitaker, students explore nonlinear regression models to explain fish weight by using fish length. They will use both transformation of the response variable and polynomial regression. Geometric interpretations of variables are leveraged to suggest nonlinear models to fit.
The intention of this lesson is for students to perform two or three linear regression analyses that feel like others they have done before. The difference is they draw on prior knowledge of geometric/physical relationships to suggest a modification to the first analysis to improve it.
Because most of the nonlinear models considered in this lesson have only a single predictor variable, students’ familiarity with simple linear regression can be extended to nonlinear modeling. If students are familiar with multiple linear regression, then two additional polynomial regression models can be included.
