Is it cheating to explore the data, because you wouldn't have registered the relevant hypothesis? Absolutely not! The idea is to have ideas! Once you get the idea you make a hypothesis & test it with a new dataset, and the whole thing repeats itself!
23,406 followers
23,406 followers
@WallaceUcsf I agree! Though in many cases it doesn't have to be a specific hypothesis, but rather a broad theoretical background about what might (and might not) be in the data. https://t.co/2PvKwW61gF
23,406 followers
@Y_Gilad I think we actually agree on this one too because the notion is just to get a new idea, not a conclusion. Once you have the new idea you can generate a new dataset to test it and the data-hypothesis conversation continues.. https://t.co/2PvKwW61gF
23,406 followers
@YannPonty Yes, you're right! That's why you have to test an idea from one dataset on a new dataset. https://t.co/2PvKwW61gF