As described. If so, how many percent of data are public leaderboard based on?
From the overall data description, it says
The dataset has been divided into three splits: train, phase-1 test, and phase-2 test
So it seems that there will be public and private LB. A safe guess is that private will be same size as public, but it would be good to get an official answer from admins.
Yes, we keep a private test dataset to evaluate the models during the second phase of the challenge. This private dataset is of the same size and same class imbalance, as the public test we delivered for the first phase.
Mosquito Alert Challenge Team.