0
$\begingroup$

I am doing regression analysis for my data , nearly half of my data is Zero .

I have conducted Generalized additive model for my data ; but I was wondering if it is enough to do only generalized additive regression analysis for my data ???

$\endgroup$
1
$\begingroup$

A large number of zeros does not necessarily mean zero inflation. It depends on the mechanism for the zeros. There are tests for zero-inflation available for the Poisson distribution (not sure about other distributions). One option is to compare the residuals from your GAM to the residuals from a zero-inflated or zero-altered model (choice depends on process generating the zeros). Are there patterns in the residuals from either model? Another option is to compare the models using AIC or similar information criterion. A third option is to compare the observed and fitted values from both models to see which of the models is doing a better job of predicting the fitted values. There are various approaches for doing this.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.