keiichishima / pcalg Goto Github PK

Hi,
I don't think the application of the rules on line 183+ is correct.
I believe these rules are supposed to be applied until the graph is no longer changing. However, you're considering each edge only once and then never checking it again.
Changing another edge later on might make it so that a rule is applicable to an earlier edge you checked, but in your case, the rule is not applied in such a case.

I was trying to use pc-algorithm use your script, but i encounter error below immediately:

File "pcalg.py", line 72, in estimate_skeleton adj_i.remove(j) AttributeError: 'dict_keyiterator' object has no attribute 'remove'
for the adj_i was G neighbors of node i(G was a graph).
it seems the adj_i was a iterator. how it could be an object of remove and len?
can you tell me how to correctly use this module?

Hi,

I want to estimate the Average Causal Effect (ACE) of X on Y using the causalEffect function. However, it seems that it only takes into account direct edges, see example below.

``
set.seed(1)

g <- randDAG(3, 2)
plot(g)
causalEffect(g, 1, 2)
``

returns the direct effect of 2 on 1 (0.4349115) instead of the total effect (0.4349115 + 0.4349115*0.615568).

Am I missing something, or is this a bug? As far as I understand, ACE is defined as E(Y | do(X = z)) - E(Y| do(X = z+1)) which should yield the total effect, as, in the example above, intervening on 2 also changes 3 which impacts 1.

Thanks for any help!

Cheers,
Fabian

I have came across this package from "Elements of Causal Inference". Thank you for your work!

I have a question, is it possible to provide a custom conditional independence test?

I generate a double/float data matrix (x_data) like the picture above, however an error occurs as below:

Is it only integer data type allowed？
Thanks!