[R-sig-eco] Zero-inflated count data and regression trees

[Andrew Halford]

Hi Listers,

I am after some help with finding a splitting criterion to deal with
zero-inflated count data in a regression tree analysis. The rpart function
only has method='poisson' which doesn't deal with excess zeroes.

I found a paper by Lee and Jin (2006) Decision tree approaches for
zero-inflated count data. Journal of Applied Statistics Vol. 33 (8)
<http://www.tandfonline.com/loi/cjas20?open=33&repitition=0#vol_33>
<http://www.tandfonline.com/toc/cjas20/33/8>:853-865,

which offers a solution using zero-inflated Poisson likelihood but I would
not know how to code this into rpart.

Any help/advice appreciated.

Andy.

-- 
Andrew Halford Ph.D
Research Scientist (Kimberley Marine Parks)
Dept. Parks and Wildlife
Western Australia

Ph: +61 8 9219 9795
Mobile: +61 (0) 468 419 473

	[[alternative HTML version deleted]]