Attempting to predict using MERF object on test dataset

[christopherDennison]

Good evening,

I have been using LongituRF to run mixed-effects random forest models, and have a question RE the predict.longituRF() function.

I have created a MERF using training data (n = 222), and want to use this model fit to predict on a subset (test data, n = 75). However, when I use the code:

predictions_merf_df <-
predict(
object = MERF_bird_Shannon,
X = data.frame(Test_list$X),
id = Test_list$id,
Z = Test_list$Z,
time = Test_list$time)

I receive the error:
Error in Ypred[w] <- f[w] + Z[w, , drop = FALSE] %*% object$random_effects[k, :
replacement has length zero

I understand this is likely not an issue with the LongituRF package itself, but perhaps something I am doing wrong on my end. I thought this would be the best place to raise the question, regardless.

Thank you so much for your time.

• Chris