### Code for AIC Multi-model analysis - Run an all model subsets analysis with multi-model inference


### import the data


datum=read.csv(file.choose())
head(datum)


### Multi-model inference uses the MuMIn package


library(MuMIn)


### Run Analysis


# First generate your global model
# if you have a lot of parameters, the 'dot' comes in very handy
results=glm(Present~.,data=datum,family=binomial) ### '.' means include all variables in datum
summary(results) ### unfortunately, we got too many variables, 'Pbird' was just used in creating data
results=glm(Present~.-Pbird,data=datum,family=binomial) ### eliminate 'Pbird'
summary(results)



### run an all subsets analysis using 'dredge' 


help(dredge)
results=glm(Present~.,data=datum,family=binomial,na.action=na.fail)
dd=dredge(results) ### only argument needed is the global model
dd ### provides an AIC table
### What if you don't want all the models?
### spit out results to file!
### Be sure to change the directory, or include path in file name!
### write.table(dd,file="results.txt")
### load package 'arm' and include argument extra=se.coef in dredge to get standard errors



### run multi-model inference using model.avg and get.models


help(get.models)
help(model.avg)
test=model.avg(get.models(dd,subset=TRUE)) ### Subset argument limits models included in analysis by AIC
summary(test)
importance(test)