Stuck at a problem with dowhy, read the docs end to end, could not find an answer. So imagine a simple graph:
X -> Y -> Z
X is present in the input data, so is Z. Y is not (unobserved). Can we get dowhy.gcm to treat such nodes as unobserved (like in dowhy.CausalModel)? When I try to auto-assign causal mechanism, the complains about Y missing from the data frame, no surprise. What is the mechanism of including unobserved variables into the causal graph with GCMs? Thanks in advance!