i was building the 3D QSAR, but q square 0.9, is there any way to validate the QSAR, so that value of q square can > 0,5?
Intan Putri I guess q2 is test set regression and r2 is training set regression, right?
Hope that your training test splitting is between 80:20 to 60:40.
Then first find the compounds in test set, which are largely distant from the test regression line.
Transfer them to training set and
transfer equal number of training set compounds (that are very close to the training regression line) to the test set.
It will reduce your r2 a bit, but q2 will increase.
repeat until your q2 is acceptable.
Uddipta Ghosh Dastidar Thank You Very Much. i will do that, very helpful answer
Please refer the articles in which authors used the biovia software for 3D QSAR.
- Badges
- Science topic