We shared some additional results here (recently updated): * [Random predictions](syn6156761/wiki/405291) * [Module sizes Round 1](syn6156761/wiki/405348) The plots on module sizes from the first round may not be that useful, especially as the background changed for Round 2, which likely affects these results. Nevertheless, we posted the results as a starting point for a discussion what aggregate information on module size / size of significant modules would be useful to share after the next round. Maybe in the end the best strategy is for each team to focus on the optimal settings for their method, but if you feel some aggregate information could be useful we would be happy to consider.

Created by Daniel Marbach daniel.marbach
> it would be useful to have some aggregate information about the significance levels of the submissions Ok, I'll think about it. I guess we could release the actual data: the p-values of the significant modules for each prediction (without telling the IDs of the modules). We'll discuss Monday, either way we would only provide additional information at the end of a given submission round. --daniel
In my humble opinion, it would be useful to have some aggregate information about the significance levels of the submissions. This could give teams more guidance about how far they are.
Dear Daniel, We would be interested in the aggregate statistics of module sizes vs overall score in subchallenge 2. We think that this would be a valuable information for many teams. Thank you
Would it be helpful to look at average significant module size too? If a team had 5000 modules but only 50 of them scored, and those 50 happened to be much larger than the others, average module size might be misleading since it factors in the sizes of the other 4950 non-scoring modules. (Correct me if I'm wrong.)   If you assume all modules in a submission are roughly the same size then average module size makes sense, but depending on the algorithm teams are using, that might not be the case.   Of course, this might all be moot since the scoring system changed, but it could be interesting to look at after the next round.

Shared results on random modules / module sizes Round 1 page is loading…