question about evaluating dataset Zi

[jesse zuo]

Hello!

The updated evaluation process says our model will be evaluated by:

reprounzip docker upload <solution> Zj.csv:Xi.csv to replace the input datasets (X2.csv, X3.csv, ...) with the hidden ones (Z2.csv, Z3.csv, ...), potentially in shuffled order (e.g., X2 can be replaced by Z3)

Is this means we should train one model based on X2 now? And this model will be evaluated using Z3 (comes from another D3 metadata). Z3 contains different attributes from X2.  So our model should have the ability to adapt different attributes from different metadata?

Thanks and regards,

Chaoji

[giovanni.simonini]

Please refer to this thread: https://groups.google.com/u/1/g/sigmod21contest/c/BQTfDnIwStc
Thanks!