Re:The Autism Phenome Project: Toward Identifying Meaningful...
by kens on Tuesday January 18, 2022 @03:06PM
(#29984953)
Very interesting! It looks like the paper examines characteristics mostly independently to form groupings. Have you considered examining all the characteristics at once using something like t-SNE to see if any new subgroups appear?
[ Parent
| on HN ]
Re:The Autism Phenome Project: Toward Identifying Meaningful...
by SubiculumCode on Tuesday January 18, 2022 @03:10PM
(#29985011)
t-SNE is a cool technique, even if some of the parameters require manual judgment/tuning. I've actually been thinking about taking our various groupings and trying a combined analysis, maybe with multivariate distance matrix regression. One issue is that labels are not exclusive, some individual may belong to several.