poi.obs.pred {COUNT} | R Documentation |
poi.obs.pred is used to produce a table of a Poisson model count response with mean observed vs mean predicted proportions, and their difference.
poi.obs.pred(len, model)
len |
highest count for the table |
model |
name of the Poisson model created |
poi.obs.pred is used to determine where disparities exist in the mean observed and predicted proportions in the range of model counts. poi.obs.pred is used in Table 6.15 and other places in Hilbe (2011). poi.obs.pred follows glm(), where both y=TRUE and model=TRUE options must be used.
Count |
count value |
obsPropFreq |
Observed proportion of counts |
avgp |
Predicted proportion of counts |
Diff |
Difference in observed vs predicted |
Joseph M. Hilbe, Arizona State University, and Jet Propulsion Laboratory, California Institute of Technology Andrew Robinson, University of Melbourne, Australia
Hilbe, J.M. (2011), Negative Binomial Regression, second edition, Cambridge University Press.
data(medpar)
mdpar <- glm(los ~ hmo+white+type2+type3, family=poisson, data=medpar, y=TRUE, model=TRUE)
poi.obs.pred(len=25, model=mdpar)