Exponential Decay Relationship in Pcount Model

[Shlomo Preiss - Bloom]

Hi All,

I am currently building and comparing 'pcount' models to fit the best one to my collected data. The relationship between the variable I am testing and recorded wolf activity seems like it might follow an exponential decay curve, but I have not found the correct way to test this in the model.

I know for quadratic/polynomial relationships the syntax is (for example):

~Temp+I(Temp^2)+I(Temp^3) etc.

However, I haven't found the right way to build an exponential decay model to compare against. I would greatly appreciate if someone could help me with building it.

Many thanks!

Shlomo Preiss-Bloom
Ph.D Candidate, Dayan Lab
School of Zoology, Tel Aviv University

[gcsadoti]

Because the log and exponent are inverse, my understanding is that one can test this (or get close) by log-transforming a covariate, i.e., log(Temp). You may have to shift Temp to deal with zeros.

[Shlomo Preiss - Bloom]

Thanks, that makes sense!

After shifting the Temp to deal with zeros, would the syntax then simply be: ~log(Temp)? 

My only other concern is that shifting the Temp (which ranges between 0 and 1) will distort its values and affect the model's integrity. Any thoughts on how to deal with this?

Thanks again!

Shlomo