Hello Solly, So, I have trained my model based on the two different phases ( time <=0 , and time <=24) . Now, as the test data-set contains gene-expression at differing time points -24,-12, 0, 12, 24 ; the model that I have designed will give probability prediction at each of these granular time points. The submission requirement however wants us to submit probability depending on the two different time phases ( <=0 and <= 24) . So, should we take an average of the probabilities that we found for each of the granular time points ( and say it represents the predicted probability of the two phases) or is there other preferred method of finding probability of the given phases. Thanks Mohammad