Should Q2 for (O)PLS-DA models be interpreted as a probability?


I was having a discussion with a collegue about the correct interpretation of Q2 value. Should it be interpreted as a probability? For exemple, if Q2 = 0.5, the class prediction will not be much better than just flipping a coin.

From my point of view, based on the formula described in Szymanska et al., Q2 is the explained variance on the predicted samples by the model or goodness of prediction. On other publication, Q2 answers for how good my model predicts a new samples. Higher the Q2, better the prediction. However, from my side the Q2 value should not be interpreted as a probability.

I would be very happy to know your opinion on this subject so that I can understand it better.

Kind regards,


Q2’s definition is very clear as explained in this post. You can see its formula from the paper referenced, which is easily interpretable and is not the same as probability

1 Like

Dr. Xia

Thank you for your prompt reply. I really appreciate the clarification of this concept.

Happy Holidays.

Kind regards.