I am writing my bachelor thesis about participation assessments with people who have had a lower limb amputation. Now I have to evaluate which of the instruments I used, offer me the most information about the client..
If you have many data derived from different instruments, you can perform a principal component analysis (PCA) on your dataset.
Let's say you have 10 instruments and 40 patients: than you can build a 40 (observations) x 10 (variables) matrix. You will get a 10x10 PC loading matrix (coeff in matlab), the first columns gives you an idea about which instrument(s) can catch the most of the patients' variation.
The matlab docs in the link below is very helpful to learn the pca basics. Read it first and then try my suggestion... let me know!
http://it.mathworks.com/help/stats/feature-transformation.html#f75476
I want to thank you for your answer, it sounds very interesting but I only have 2 instruments and 5 patients: you think I can use the PCA on this?
Mmm pca is meant to reduce data dimensionality, so it hardly fits on a small sample size dataset.
Are you measuring the same variable with different instruments? If so, try to give a look to the bland altman method or try with non parametric statistics (i.e. mann-whitney test) between the two groups.
Article International standards for early fetal size and pregnancy d...
Yes, I am indeed measuring the same variable.. I just read a little part of the article and want to thank you for helping me out with this. It looks easier than it sounds so I certainly will try to use this.
- Badges
- Science topic