I have question about building 3DQSAR model via regression model. I cant achieve r^2 more than 0.6 for 'test set' as well as for 'training CV'-what is this?. I am wondering whether 'training set' should be as similar to each others as its possible and use 'reference cmpd' belonging to the 'training set'? Or 'reference set' may be different compound - not similar to training compounds but active.

More Damian Kułaga's questions See All
Similar questions and discussions