Christoph Helma wrote:
> Excerpts from Nina Jeliazkova's message of Wed Mar 17 10:13:55 +0100 2010:
>   
>> Hello All,
>>
>> Here is an initial page for applicability domain and an API proposal,
>> based on a discussion with Pantelis (NTUA).
>>
>> http://opentox.org/dev/apis/api-1.1/ApplicabilityDomain
>>
>> Something to discuss next Monday perhaps.
>>     
>
> Looks good at a first glance - will try it out with one of the next
> updates. Models that use an external AD algorithm would have to
> "recommend" an AD algorithm - maybe this information should be added to
> the model representation.
>   
Agree. 

Going one step further,  what about introducing somehow "types" of
applicability domain methods, which might be suitable for particular
predictive method?

Best regards,
Nina
> Martin: Do you think that the proposal is useful/sufficient for the
> validation service - it is IMHO crucial to validate AD models (i.e.
> prove that predictions within (close to) AD are more accurate than
> predictions out of (far from) the AD).
>
> Best regards,
> Christoph
> _______________________________________________
> Development mailing list
> Development at opentox.org
> http://www.opentox.org/mailman/listinfo/development
>