PCA repeated measures

Discuss statistics related things
Post Reply
EDL_
Posts: 2
Joined: Mon Feb 20, 2023 2:21 pm

PCA repeated measures

Post by EDL_ »

Hi all,

Is there a way in Jamovi to produce a biplot from a PCA while taking into account repeated measures?

My dataset has a long type structure (each ID is repeated 5 times - one for each observation).

I need to perform a PCA and I have tried with both MEDA and snowCluster, but when I enter my Individual Label I receive error messages (e.g., non convenient data or duplicate not allowed).
How can I account for the multiple observations for each participant?

Thanks a lot in advance
User avatar
Arnaud Mortier
Posts: 32
Joined: Tue Apr 27, 2021 11:39 am

Re: PCA repeated measures

Post by Arnaud Mortier »

Hi,

No matter what I do in MEDA, I also get the message "non convenient data", and this thread is the only place that Google knows about where the error message was reported.
I'm using Jamovi 2.3.17.0 and MEDA 1.0.0 on Ubuntu 22.04.
I've tried everything I could, different files, different ways to fill in variables.
Screenshot from 2023-03-08 16-44-19.png
Screenshot from 2023-03-08 16-44-19.png (19.77 KiB) Viewed 4053 times
Thanks for any help.
Best
Arnaud
User avatar
jonathon
Posts: 2613
Joined: Fri Jan 27, 2017 10:04 am

Re: PCA repeated measures

Post by jonathon »

if you're having trouble with MEDA, it might be worth reaching out to the module's author.

jonathon
sle
Posts: 9
Joined: Mon May 07, 2018 2:33 pm

Re: PCA repeated measures

Post by sle »

You should have a look at this link:
https://sebastien-le.github.io/medasite/PCA.html#Potential_issue(s)

As the program gives you an automatic description of the 5 first dimensions, you should at least have 5 variables in your PCA. If you have less than 5 variables, you have to go to the numerical indicators menu and change the number of dimensions you want to describe automatically.
EDL_
Posts: 2
Joined: Mon Feb 20, 2023 2:21 pm

Re: PCA repeated measures

Post by EDL_ »

Thanks for your replies. I actually have 10 active variables and one supplementary cataegorical variable (4 levels). I can perform the PCA in Jamovi only if I do not enter the ID variable. As soon as I add it I get the error message (non convenient data).
In the end I solved the issue by changing software so I cannot be much of help.
Post Reply