Article ID: 264 - Last Modified: May 9, 2011
What is the meaning of increasing R-squared and Q-squared values with increasing number of PLS factors?
R-squared will always increase as you add more PLS factors because it measures the strength of the least-squares fit to the training set activities. More precisely, an R-squared value of 0.9 means that the model accounts for 90% of the variance in the observed activities for the training set. The value gets closer and closer to 1 (i.e., 100%) as more PLS factors are incorporated into the fit.
Q-squared is an analogous statistic, except that it comes from applying the QSAR model to the test set. Since the model is not directly calibrated to fit the test set, Q-squared may or may not increase as you add more PLS factors. But if it's a good model (i.e., it embodies the SAR and you've picked a reasonable number of PLS factors), Q-squared will be comparable in value to R-squared.
Keywords: Phase, QSAR, PLS factors, R-squared, Q-squared
Type the words or phrases on which you would like to search, or click here to view a list of all
Knowledge Base articles