glm - R Zeroinfl model -
glm - R Zeroinfl model -
i carrying out zero-inflated negative binomial glm on insect count info in r. problem how r read species info 1 stacked column preserve 0 inflation. if subtotal , import r single row titled abundance, loose zeros , model doesn't work. already, have tried to:
stack info myself (there 80 columns * 47 rows) 3760 rows after stacking manually can imagine how slow r gets when using pscl zeroinfl()
command (it takes 20mins on computer!, still worked)
the next problem concerns spatial correlation. samplers sampled same medium violate independence. can set medium in factor in model?
3760 rows take 20 mminutes pscl? god, have battle 30.000 rows :) that´s why pscl calculation did not finish...
however, worked glmm including nested random effects (lme/gamm) , negative binomial distribution setting theta low value distribution handles 0 inflation. think depends on grade of zeros. in case 44% , residuals looked rather good.
r glm zero
Comments
Post a Comment