Another scoring scheme based on BDeuScore

cmu-phil / tetrad

Repository for the Tetrad Project, www.phil.cmu.edu/tetrad.

GNU General Public License v2.0

404 stars 111 forks source link

Another scoring scheme based on BDeuScore #232

Closed haluk closed 8 years ago

haluk commented 8 years ago

I need to define another scoring scheme based on BDeuScore but it has little bit different formula. What scoring method does Fgs use if i set BDeuScore as a scoring method. localScore method inside the BDeuScore has different definitions. Which one should I override to perform Fgs?

jdramsey commented 8 years ago

The one with target index and int[] parent indices. All the other ones can point to that.

On Wed, Jul 13, 2016 at 7:00 PM, Haluk Dogan notifications@github.com wrote:

I need to define another scoring scheme based on BDeuScore but it has little bit different formula. What scoring method does Fgs use if i set BDeuScore as a scoring method. localScore method inside the BDeuScore has different definitions. Which one should I override to perform Fgs?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/cmu-phil/tetrad/issues/232, or mute the thread https://github.com/notifications/unsubscribe/AJZZRwxnB4_o9NOSm2VoF2qmfE-1qGGoks5qVW4pgaJpZM4JL7pY .

Joseph D. Ramsey Special Faculty and Director of Research Computing Department of Philosophy 135 Baker Hall Carnegie Mellon University Pittsburgh, PA 15213

jsph.ramsey@gmail.com Office: (412) 268-8063 http://www.andrew.cmu.edu/user/jdramsey

haluk commented 8 years ago

Do you use scoreDag method in FGS to compare candidate graphs?

jdramsey commented 8 years ago

I don't, but you can. That is, different DAGs over the same variables and dataset. It returns a BIC score, which is comparable only up to a constant, the constant being determined by the dataset (and its variiables).

On Mon, Jul 18, 2016 at 5:19 AM, Haluk Dogan notifications@github.com wrote:

Do you use scoreDag method in FGS to compare candidate graphs?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/cmu-phil/tetrad/issues/232#issuecomment-233281801, or mute the thread https://github.com/notifications/unsubscribe-auth/AJZZRwO3UwyjAn_hVPP2KOc5V161NIyAks5qW0U0gaJpZM4JL7pY .

Joseph D. Ramsey Special Faculty and Director of Research Computing Department of Philosophy 135 Baker Hall Carnegie Mellon University Pittsburgh, PA 15213

jsph.ramsey@gmail.com Office: (412) 268-8063 http://www.andrew.cmu.edu/user/jdramsey

haluk commented 8 years ago

How about if I set different scoring scheme? Then it won't calculate BIC score right?

jdramsey commented 8 years ago

?? Read up on BIC scores. Read Schwarz's article. Not just any score will work.

On Mon, Jul 18, 2016 at 5:49 AM, Haluk Dogan notifications@github.com wrote:

How about if I set different scoring scheme? Then it won't calculate BIC score right?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/cmu-phil/tetrad/issues/232#issuecomment-233287589, or mute the thread https://github.com/notifications/unsubscribe-auth/AJZZR00lbgUuI3GpHkfynYsd33qcqWbmks5qW0xAgaJpZM4JL7pY .

Joseph D. Ramsey Special Faculty and Director of Research Computing Department of Philosophy 135 Baker Hall Carnegie Mellon University Pittsburgh, PA 15213

jsph.ramsey@gmail.com Office: (412) 268-8063 http://www.andrew.cmu.edu/user/jdramsey

jdramsey commented 8 years ago

Sorry, no, if you set a different score it will calculate that score. (Did I miss you question?)

Check out the Chickering 2002 paper--you need a consistent, decomposable score. Chickering gives the conditions.