Hello,
I have recently installed proc lca and i have tested with a small dataset and everything seems to work pretty fine. However I need to run such analysis for longitudinal data (approximately 32million individuals overall). I have been trying to test the different models with a pretty small sample (of 100000 individuals), but above that it starts giving me issues. My aim was to run proc lca on 10% of my overall population and afterwards assign people to the clusters found in such sample. But it gives me this errorr of overflow detected lca. Does anyone know what could I do? Also what is it the best seed in this case with such a big population?
This is the current basic model I am testing:
proc lca data=test;
title "6 class model";
nclass 6;
items contract temporary multjobs income catcba;
categories 5 2 3 5 3;
seed 100000;
groups sex;
RHO PRIOR=1;
run;
everything seems to work when I run it with groups or covariates etc. but until 100000 individuals.
Thanks you!