Is there a way to read a fit.mle item (for comparison with anova) into R?

Question

I've fitted 9 different models to my data using the R package diversitree, which returns a list of class "fit.mle", which contains estimated parameters and log likelihood. I'd like to run anova to compare the models. Normally I estimate all 9 models and compare them, but my current data is so big I need to estimate each and write to file before comparison. How should I go about running anova on these?

Answer 1

From the package documentation I can see that {diversitree} includes an S3 anova() method for objects of class fit.mle.

The syntax would be anova(model1, model2, model3) where model1, model2, model3 have class fit.mle.

You can do something like:

# ... fit model1

# write model1 to rds file
writeRDS(model1, "model1.rds")

# ... fit other models and write each to file in new sessions as needed

# read each model from rds
model1 <- readRDS("model1.rds")
model2 <- readRDS("model2.rds")
model3 <- readRDS("model3.rds")

# compare model1, model2, model3 using `diversitree::anova(<fit.mle>)`
anova(model1, model2, model3)