Hi Solly, I have a question. There is 166 set of genetics data at different times for only 23 unique SUBJECTIDs, and the questions are predicting output variables at time zero and 24 (using genetics data up to zero and 24, respectively), correct? The problem is that we don't have genetics data at time 0 or 24 for most of SUBJECTIDs in testsets. So, somehow the model has to Predict the output at time 0 and 24? Please let me know if I didn't get the questions. Thanks,

Created by shamim sarhadi ssarhadi
Up to and including. Please re-read my first response for details on which data may be used for the two different sets of predictions. I've explained it in detail there.
Thanks Solly but " up to 0 & 24" or EXACTLY at 0 & 24?
Shamim- For each Subchallenge in which you participate, you are allowed 4 submissions to the leaderboard. One must be your official `Up to Time 0`, and one must be your official `Up to Time 24` submissions. The remaining two may be test submissions used at you discretion. In other words, your two timepoint submissions are submitted as separate files, and you must indicate which is which by naming the submission appropriately and documenting the Synapse ID of the submitted file in your writeup.
Thanks Solly! Now it is crystal clear that for building our predictor models we have to use data in 2 different time ranges, i.e. Up to Time 0 and Up to Time 24, for all three subchallenges, i.e. different outcomes including Shedding, Symptomatic and LogSymptomScore. The only part which is not clear to me is the predictions for testset which we submit to leaderboard for scoring. In testset there is more than one set of genetics data for every single SUBJECTID, which is correspond to genetics data at different times. On the other hand, we have to predict ONLY one output for every single SUBJCETID. The question is, at what time we have to predict? If we are using genetics data "up to time 0, our prediction for every single SUBJECTID have to be the predicted output at time 0?, and for using genetics data "up to 24 h" our prediction for every single SUBJECTID have to be the predicted output at time 24 h? Thanks!
Shamim- What you're discussing are the timepoints, not the subchallenges (which refer to the 3 different outcomes: Shedding, Symptomatic and LogSymptomScore). For each subchallenge you participate in, you're required to submit a set of materials using data in 2 different time ranges: `Up to Time 0` and `Up to Time 24`, where the former corresponds to any of the predictors from the files in the `Phase 1` folder and the latter corresponds to any of the predictors from the `Phase 1`, `Phase 2`, and `Phase 3` folders. In other words, you can use any gene expression at any timepoint up to and including time 0 or up to and including time 24 for the two time ranges. I hope that's more clear.

clarification of subchallenges' questions page is loading…