Hi!
I am a PhD student in my third month with a background in cognitive neuroscience. I loved statistics and thus developed a rather deep understanding of regression analyses.
Now, I started to apply computational models/analyses to the data I am currently looking at and the next step involves using PCA. My supervisor is talking a lot about how interesting it would be to see the component's eigenvalues.
While trying to get my head around those, I found that they sound quite similar to what semipartial correlation coefficients in (linear) regression analyses express. Is that right? Can they even be viewed as being equivalent? Or am I misunderstanding something fundamental?
I hope this is the right place to ask questions like this, I was not able to find anything that appeared to be more suitable.
Thank you very much for any help!
Are PCA eigenvalues equivalent to semipartial correlation coefficients in linear regression?
Moderators: AmyCowen, kgudger, bfinio, MadelineB, Moderators
-
- Moderator
- Posts: 908
- Joined: Fri Jun 20, 2014 4:42 pm
- Occupation: Biostatistician/Data Scientist
- Project Question: Interested in volunteering as an expert
- Project Due Date: n/a
- Project Status: Not applicable
Re: Are PCA eigenvalues equivalent to semipartial correlation coefficients in linear regression?
Hello,
Science Buddies is for students in grades K-12 who want help with their science projects.
So, no, this is not the appropriate forum for your question. I would suggest that you look for course (on line or on the ground) on PCA, so that you can develop a better understanding of that topic. When you have specific questions, you might try stats.stackexchange.com.
Thank you.
Science Buddies is for students in grades K-12 who want help with their science projects.
So, no, this is not the appropriate forum for your question. I would suggest that you look for course (on line or on the ground) on PCA, so that you can develop a better understanding of that topic. When you have specific questions, you might try stats.stackexchange.com.
Thank you.