I have a few questions regarding the ranking of the submissions for intermediate and final rounds:
1- It was mentioned that we do not have to submit predictions for subjects with no training labels. Does it mean that we can exclude them from our submitted CSV file? Or should we keep the measurement_id and just put zeros in the prediction column?
2- What is the null model that the submissions are compared with?
3- There are some files in the testing folders with only a few data points (e.g. 19b9483f-fb7e-4fad-abe6-de00fb7e1331.csv contains only 122 seconds of data). Are they also included in the comparisons?
Created by Nader Naghavi ndrngh @ndrngh -
1) You have been provided [submission templates](syn21348877) containing all of the measurement_ids which will be scored. This should help you identify all the measurement_ids for which predictions are expected. Note that the files differ by subchallenge, so please be sure to use the correct one.
2) This question has already been answered on the discussion forum: https://www.synapse.org/#!Synapse:syn20825169/discussion/threadId=6925
3) See (1). The minimum requirement for inclusion was 1 minutes worth of active data.