@yacc143
I bet if the lines on the graph broke each time the math/definition/diagnostic criteria changed there'd be less confusion.
dsm4 autism: red
dsm5 autism: yellow
"What's causing this?!" Umm, the dsm5 aka the slow grind of human knowledge
Just like there's more adhd now because we literally didn't call it that before... it was ADD or hyperwhatever or "just being a spacy kid, man" or "could do better if they just applied themselves".
Even a vertical rule on the graphs for each revision of the book would provide some clue, but knowning which version the study used would be nice
@yacc143
> practices were totally divergent.
> So I'm totally confident in numbers being absolutely comparable.
How's that work? The total percentage of the population its the same but the division of categories differs?
@EndlessMason @ichbindabomb @isotopp I have no idea, that was over a decade ago, and my focus and knowledge about that stuff back then wasn't as intense as say 5 years ago.
But as a guess, validating the classification systems (our first "complete" classification system was an AU one, and hence we had pre-classified training data from AU, while surprise, the first paying and important customers were from the US/UK).
@yacc143
The classification of papers? Or of humans?
@EndlessMason
Of texts as such. But obviously if your definition of diagnosis X does not match my definition of diagnosis X, than we will qualify patients differently.
Because of we use your definition of X, then suddenly my bibliography on topic X is only classified as X at 10% hit rate, than obviously the classifier dies not think that my bibliography is about X at all. Not your X at least.
Schneewittchen
Andreas K
EndlessMason