Timeline for "Balancing" principal components
Current License: CC BY-SA 3.0
4 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 14, 2015 at 11:27 | comment | added | user78229 | @Gardner Are you using the right words? "Dimensions" don't load on dimensions, input variables load on dimensions. If you are saying that, after rotation, the loadings for your input variables were maximized on a single dimension and zero (or close to it) on the other dimensions, then that would be the "simple structure" that is sought when rotating. | |
| Jun 14, 2015 at 10:23 | comment | added | vigji | Thank you very much for your kind explanation. I have almost solved my problem with data normalization. Just another question, for the sake of curiosity: I tried to "rotate" the pc matrix, and my output was not exactly what I was expecting: each one of the new pcs had a strong value on only one original dimension, and was almost 0 on each of the other dimensions. Is there a way to perform a similar transformation, giving the constraint that "the first two load should be maximized"? | |
| Jun 13, 2015 at 14:56 | vote | accept | vigji | ||
| Jun 13, 2015 at 12:10 | history | answered | user78229 | CC BY-SA 3.0 |