Created by Eiman Ahmed eahmed Hi @Eiman ,
Yes, you will be able to test your model through the submission system during the Synthetic Validation Phase. Please note that the purpose of this phase is **for testing your model only**, e.g. you will know if it is VALID or INVALID. Scores will not be returned.
Best,
Verena Dear @Michael.Mason
Thanks for your clarfication.
One more question please, so during synthetic validation phase, participants will be able to test their models in the synthetic data through the submission system?
Thank you again, Dear @Eiman ,
This will be covered during the webinar but I am happy to explain a bit here. The Challenge is unique in that it does not provide training data or intermediate testing data. This is due to the fact that the validation data is unique (clinical trial data of an immune checkpoint inhibitor) and large enough for validation but not large enough to split up into intermediate testing let alone splitting into validation *and* training data. To allow people to test the functionality of their algorithm we provide synthetic data: a subset to facilitate data ingest locally and a larger dataset in the submission system. Because the real validation data is such a highly valued resource the synthetic data *does not* mimic any of the real variable distributions and will tell you nothing about the accuracy of your model. The synthetic data section of the [data sub-wiki](https://www.synapse.org/#!Synapse:syn18404605/wiki/607227) has some detail on this.
Cheers,
Mike Dear @Michael.Mason
can you please explain about the phases (synthetic validation and validation phases) and how it goes. It's quite unclear to me.
Thank you,