How can distance correlation be used to develop a model?

37 Views Asked by At

I've been reading up on distance correlation recently; it's being billed as a way of testing if two variables are related in a non-linear fashion. Now, for a (non-zero) standard correlation coefficient, you know that a linear model will capture some of the behavior. As far as I can tell, there is no similarly straightforward way to figure out the form of the model you should be using for two variables that have non-zero distance correlation. Is there such a way, or do you have to break out your other model-building techniques?