Donate data before May 17

[Isabelle]

You have until Friday May 17 to contribute your own pairs and enter the challenge in track 1.

This is EASY: just format a few pairs in the data format of the challenge and email them to causality at chalearn ... org. And you may win one of the track 1 best paper awards for proposing an interesting dataset!

It is VERY important that a lot of people submit data, even artificial data. This will considerably strengthen the validity of the results of the challenge by lowering the risk that the algorithms proposed are capitalizing on artifacts of the proposed dataset. We have already become aware of a few artifacts that will be corrected in test data, such as uneven distributions of number of values across classes, which will be corrected in the test data. But it is very difficult to assertain that the data will be completely devoid of artifacts. You may contribute:

- real data with known causal relationships

- real data with unknown causal relationships (hence there is a potential for making new interesting scientific discoveries if there is a consensus between methods)

- artificial data.

There is no mimimum number of pairs nor minimum number for samples, but we recommend submitting at least 10 pairs and 100 samples per pair. The data you provide will not be used for the final ranking, but, we will publish the results of the best ranking methods on those data for comparison. 

[Hari Koduvely]

Hi Isabelle, I know it may be too late into competition to suggest this. I have been using some of the data sets from the Bayesian Network Repository for my research. May be this would of use for this competition. http://www.bnlearn.com/bnrepository/ 

I noticed about this competition in Kaggle only very recently.

Thanks and Regards,

-Hari