Although we have focused on linear regression in this chapter, you certainly are not limited to performing regression with linear formulas. You can model your dependent variable by one or more nonlinear terms such as powers, exponentials, or other transformations on your independent variables. For example, we could model Sales by a polynomial series of TV terms:
Keep in mind, however, that as you add this complexity, you are again putting yourself in danger of overfitting.
In terms of implementing non-linear regressions, you cannot use github.com/sajari/regression, which is limited to linear regression. However, go-hep.org/x/hep/fit allows you to fit or train certain nonlinear models, and there are other various people in the Go community that have, or are, developing other tools for nonlinear modeling.
There are also other linear regression...
