Greedy TT structures learning

[frito]

when I submit continuous data to get a structure, are the data first discretized by Genie or the score calculation is done on the original continuous data?

Is the local score comparison accomplished simply by e.g. score_old<score_new (hence choosing the new model here), or some significance test is performed?

thx

[shooltz[BayesFusion]]

when I submit continuous data to get a structure, are the data first discretized by Genie or the score calculation is done on the original continuous data?

If any data column selected for learning is continous, GeNIe will not run GTT - you'll get the "Selected algorithm doesn't support continous columns" message.

[frito]

when I submit continuous data to get a structure, are the data first discretized by Genie or the score calculation is done on the original continuous data?

If any data column selected for learning is continous, GeNIe will not run GTT - you'll get the "Selected algorithm doesn't support continous columns" message.

Oh I made a mistake I ment the PC search, and since it uses a cond independence test I suppose genie does NOT discretize the data ?

Also what indep. test it uses Chi^2?

How is it possible it is so incredibely FAST, other PC implementations I used for 100 variables simply did not work at all some resulted in two arcs for 100 variables and other implementations required exponential space(so could not run)?

thx

edit:
I posted this into a new post with appropriate name
http://genie.sis.pitt.edu/forum/viewtopic.php?p=705#705