bd2kccd / causal-cmd

16 stars 8 forks source link

Use the tetrad software #105

Open miaomiao-alt opened 1 month ago

miaomiao-alt commented 1 month ago

Hello author, I have 58 continuous variables, want to find the causal relationship between these variables, first I use the BOSS algorithm (using the default parameters), and set the level constraint (level 0 has 47 variables, level 1 has 11 variables), and then I get the following graph, I do not know how to deal with the undirected edge in the result. As a result, when setting the SEM parameter model in the next step, the undirected edge in the above results prevents me from carrying out this step, I would like to ask you how to solve it.

4
jdramsey commented 1 month ago

Hi @miaomiao-alt! Maybe this should be posted to a Tetrad issues list instead of the causal-cmd issues list, but let me answer it.

You have a graph that looks like it's a CPDAG. (Perhaps you generated this using BOSS, e.g.). For estimation, you need a DAG. The thing to do is put another Graph box on the workbench and draw an arrow from your graph box (or search box) to this new graph box, double click the new graph box, and select Choose Dag in CPDAG). This will give you a DAG. Then you can make a SEM PM with the DAG or estimate it.

You may also want a particular DAG in the CPDAG. For that, maybe post the issue to the Tetrad issues list, https://github.com/cmu-phil/tetrad/issues, and I'll give you step by step instructions.

Hmm.. come think of it, maybe there should be a gadget in causal-cmd to do this. I'll ask.

Best, Joe

miaomiao-alt commented 1 month ago

OKOK,thank you so much!I will post the issue to the Tetrad issues list.