The evaluation for the competition will be counted based on the best prediction **accuracy** resulted by the model