NeurIPS 2022 CausalML Challenge: Causal Insights for Learning Paths in Education Forum

Go back to competition Back to thread list Post in this thread

> Lower triangular adjacency matrices?

The adjacency matrices in the local development set for Task 1 (and I believe also in the public leaderboard set) are all lower triangular. Can we assume the same for the private leaderboard set? This would make discovery easier.

Posted by: bbussmann @ July 4, 2022, 1:06 p.m.

Dear participants,

Thanks for mentioning it, we also realise this, which allows decent score even with trivial solutions. To remedy this, we will re-generate the data JUST for private leaderboard.

We will send a notification to all participants once the private data are ready to download, which should be within in a few days.

This means you can continue to develop the model using local_dev and public, but shouldn't assume that the adjacency matrix is lower triangular.

We are so sorry for the inconvenience.

Posted by: WenboGong @ July 5, 2022, 4 p.m.

Hello, I want to ask if you have accomplished the re-generation of the private dataset?

Posted by: skt @ Sept. 7, 2022, 1:53 p.m.

Hi,

We already finished re-generating the private dataset back on 07 July and sent reminder emails to all participants about this regeneration.

I just want to confirm whether you have successfully received the email or not?

Posted by: WenboGong @ Sept. 7, 2022, 2:04 p.m.

Thank you for the reply! I have not received the email. But I am not sure if other participants received the email or not.

Posted by: skt @ Sept. 7, 2022, 2:40 p.m.

Hi,

We want to double-check if you have received the reminder email about the opening of task 3 and 4?

Thanks

Posted by: WenboGong @ Sept. 7, 2022, 2:48 p.m.

I also did not receive it. Maybe it was recognized as trash mail, and I have not noticed it.

Posted by: skt @ Sept. 7, 2022, 3:07 p.m.
Post in this thread