Model Browser User's Guide

Transforms

The following example shows the transforms available.

Input transformation can be useful for modeling. For example, if you are trying to fit

using the log transform turns this into a linear regression:

Transforms available are logarithmic, exponential, square root, , 

, and Other. If you choose Other, an edit box appears and you can enter a function.

Apply transform to both sides is only available for nonlinear models, and transforms both the input and the output. This is good for changing the error structure without changing the model function. For instance, a log transform might make the errors relatively smaller for higher values of x. Where there is heteroscedasticity, as in the Covariance Modeling example, this transform can sometimes remove the problem.

Covariance Modeling

Designing Experiments