I am using the effects package to construct some probability graphs showing the predicted probabilities from a logistic regression model.However, I get an odd error message and don’t know what the issue is.
When I attempt to generate the plots, I get the following error. The warning is not an issue, it’s that I’m not understanding what the error message is telling me.
library(effects)
dat$won_ping = as.factor(dat$won_ping)
mod2 = glm(won_ping ~ our_bid +
age_of_oldest_driver2 +
credit_type2 +
coverage_type2 +
home_owner2 +
vehicle_driver_score +
currently_insured2 +
zipcode2,
data=dat, family=binomial(link="logit"))
> plot(effect("our_bid*vehicle_driver_score", mod2), rescale.axis=FALSE, multiline=TRUE)
Warning message:
In analyze.model(term, mod, xlevels, default.levels) :
our_bid:vehicle_driver_score does not appear in the model
Error in plot(effect("our_bid*vehicle_driver_score", mod2), rescale.axis = FALSE, :
error in evaluating the argument 'x' in selecting a method for function 'plot': Error in apply(mod.matrix[, components], 1, prod) :
subscript out of bounds
Here’s info on my data and my glm commands:
> str(dat)
'data.frame': 85240 obs. of 71 variables:
$ our_bid : num 155 123 183 98 108 159 98 123 98 200 ...
$ won_ping : Factor w/ 2 levels "0","1": 1 1 2 1 1 1 1 1 1 1 ...
$ zipcode2 : Factor w/ 4 levels "1:6999","10000:14849",..: 4 3 2 1 3 2 3 1 2 2 ...
$ age_of_oldest_driver2 : Factor w/ 4 levels "18 to 21","22 to 25",..: NA 3 NA NA NA NA 3 NA 3 NA ...
$ currently_insured2 : Factor w/ 2 levels "0","1": 2 1 2 2 1 1 2 2 1 1 ...
$ credit_type2 : Ord.factor w/ 4 levels "POOR"<"FAIR"<..: 2 3 2 3 2 2 1 3 3 2 ...
$ coverage_type2 : Factor w/ 4 levels "BASIC","MINIMUM",..: 4 3 3 3 3 3 3 3 4 3 ...
$ home_owner2 : Factor w/ 2 levels "0","1": 1 2 2 2 2 2 2 2 2 2 ...
$ vehicle_driver_score : Factor w/ 2 levels "0","1": 1 1 1 1 1 1 1 1 1 1 ...
And finally, here might be some useful info:
> sessionInfo()
R version 2.14.0 (2011-10-31)
Platform: x86_64-pc-mingw32/x64 (64-bit)
locale:
[1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252 LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C LC_TIME=English_United States.1252
attached base packages:
[1] grid stats graphics grDevices utils datasets methods base
other attached packages:
[1] effects_2.2-1 colorspace_1.1-1 nnet_7.3-1 MASS_7.3-16 lattice_0.20-0 foreign_0.8-46
loaded via a namespace (and not attached):
[1] tools_2.14.0
Help! What is the error message mean? Normally, if a “subscript is out of bounds” that’d mean I’m selecting something outside the bounds of that data structure, but that simply is not occuring.
EDIT:
To @Rowland
As I said above, the warning and error messages are seperate and unrelated. Let’s say I take out zipcode2 and run the glm:
mod2 = glm(won_ping ~ our_bid +
age_of_oldest_driver2 +
credit_type2 +
coverage_type2 +
home_owner2 +
vehicle_driver_score +
currently_insured2,
data=dat, family=binomial(link="logit"))
> plot(effect("our_bid*home_owner2", mod2), rescale.axis=FALSE, multiline=TRUE)
Warning message:
In analyze.model(term, mod, xlevels, default.levels) :
our_bid:home_owner2 does not appear in the model
This produces just the warning, which is fine as I get the desired result. So the fact that “:” does not appear in the model is not the issue, and DOES NOT cause the error message.
Try this:
I suspect you have some unpopulated cells. With your edit, it seems unlikely that the sparseness I am hypothesizing as the problem lies with these two variables. It still remains possible that despite your large number of cases that there are still empty cells when the model is constructed with all of those factor variables.