Some puzzles about model selection and "Mclust could not estimate model 2 with 4 classes".

[Yifei Wu]

Dear all, 

I would like to group people into 3-6 classes based on three variables. I'am a little bit confused about model selection. According to the tutorial "package 'tidyLPA' ", we can select the best model through analytic hyrarhy process (AHP). For example, AHP suggests me to use model 4, as shown in the picture.

However, the results showed that "Best model according to BIC is Model 6 with 3 classes."when I run following codes to compare solutions. So I am not sure which model I should select or maybe I understand wrong.

library(tidyLPA)

library(tidyverse)

HKG %>%

    select(BSBGSLS, BSBGSCS, BSBGSVS) %>%

    scale() %>%

    estimate_profiles(3:6, models= c(1,2,3,6)) %>% 

    compare_solutions(statistics = c("AIC", "BIC"))

results:

In addition, I've deleted all missing values in my dataset before I use tidyLPA. I haven't found the reason for so many warning messages.

Hope you can help me figure out and thanks a lot.

Kind regards

Yifei