Hi all,

there are new features to the validation reports.

First, a confidence plot (could be named "applicability domain curve" as
well) shows how the prediction confidence effects accuracy/rmse.
See (classification):
http://toxcreate2.in-silico.ch/validation/report/crossvalidation/44
and (regresssion):
http://toxcreate2.in-silico.ch/validation/report/crossvalidation/53

Second, there is a report to compare different algorithms/parameter settings
that were used on the same dataset (including significance tests).
See (classification):
http://toxcreate2.in-silico.ch/validation/report/algorithm_comparison/48
and (regresssion):
http://toxcreate2.in-silico.ch/validation/report/algorithm_comparison/55

The latter can be used to compare the efforts of different groups on the
REACH endpoint prediction. Please note that the results (especially
regression) are far from good, this is just a proof of concept (I have to
re-check the r² value computation as well).

How do you like it? Any comments, feature requests?

Regards,
Martin

P.S.
The current validation development version is running at
http://toxcreate2.in-silico.ch/validation, http://
opentox.informatik.uni-freiburg.de/validation will be updated soon.

-- 
Dipl-Inf. Martin Gütlein
Phone:
+49 (0)761 203 8442 (office)
+49 (0)177 623 9499 (mobile)
Email:
guetlein at informatik.uni-freiburg.de