Dear Organizers, Initially, I would like to thank you for giving an opportunity to work on real-life biological data. After the completion of round 1, I would like to know: What thresholds of RMSE/Scores would be practical for scientific usage? It is obvious that prediction scores can not be zero (owing the noise in the data). Best Regards, Krishna

Created by krishna singh krishna14
Hi Krishna, It's good to see that you enjoy working on this dataset. In all subchallenges, we have a score for "random noise". Anything lower than that captures some signals. It is very nice to see, that (almost) all submissions are better than random. I personally dont know what is the lowest bound of the RMSE on this data and I am not sure if such bound can be established. Even though the models won't be perfect, I think we will learn something about what they can actually predict and in which situations they fail. That's already a scientific result. Maybe someone can add more on this.:) best, Attila

Question to scientists/organizers page is loading…