@slatestarcodex They're doing a regression targeting ABIDE's measured IQ scores, so I guess that would fall under 'retrodiction'?
-
-
-
@gwern I meant if they tested it on the same data they derived it from. Fakey's saying they didn't. -
@slatestarcodex Ah, whether they overfit? Does Fakey object to the crossvalidation procedure they used or something? -
@gwern No, I'm interpreting Fakey as saying it's legit. -
@slatestarcodex Then should be fine. Dataset's public anyway, so can always extract estimated models and see how they perform on future data -
@gwern@slatestarcodex Would the dataset work for doing what I'm trying to do here? http://emilkirkegaard.dk/en/?p=5034 -
@KirkegaardEmil@slatestarcodex I'm afraid you're more qualified to answer that question than either of us!
End of conversation
New conversation -
-
-
@slatestarcodex@maj_psy I think it's legit but the age range (6-15) iq diffs means you're essentially predicting brain maturation. -
@slatestarcodex@maj_psy would be surprised (but interested!) if they get anything close in age homogeneous adults. I predict probably not
End of conversation
New conversation -
-
-
@slatestarcodex they're using 10-fold cross-validation, which means more or less prediction -
@slatestarcodex they chop the data into 10 parts and then for each part train their model on the other 9 parts and predict the 10th -
@slatestarcodex not as good as predicting a fully novel data set but enough to often not be bullshit
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.