Is there a guarantee that we will be tested ,in the private test set ,on the same target distribution ( ratio = 85/15 ) than the public one? or is there a chance the distrubution is different? let's say 50/50 for example?
It might be same or different. we never can tell. It depends on how the competition organisers prepared the dataset