I am running a two component PLS model, and I am interested in the Q2 of the individual Y variables (not the average or total Q2 of the model).
My question is: how to interpret results when the Q2 of the first component is positive and the Q2 of the second component is negative; or, when the Q2 of the first component is negative and the Q2 of the second component is positive.
Consider the following four possibilities of Q2 values for a single Y variable in a 2-component PLS model:
|Q2 of Comp1|Q2 of Comp2|
| 0.5 | -0.3 |
| -0.3 | 0.5 |
| 0.3 | -0.5 |
| -0.5 | 0.3 |
Would we consider the model “predictive” in any of these situations?
Any insight would be greatly appreciated!