You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
Causal graphs and their implementation inherently affect all upstream causal inference procedures. Most importantly dodiscover relies on the API of pywhy-graphs. Any changes introduced here should not break the "next version of dodiscover". In order to catch these, we need integration tests.
Describe the solution you'd like
There are three options:
integration tests in pywhy-graphs
integration tests in dodiscover
integration tests in both
I am leaning towards adding this to dodiscover only, or both. We can delete one of them down the line, but the CI times are already pretty fast, so we can sacrifice this for the sake of ensuring integrability across the two packages (and eventually dowhy).
The solution would just be to create a tight "accuracy" constraint on causal discovery algorithms that leverage each graph class and then this performance should not change across any changes done in pywhy_graphs.
The text was updated successfully, but these errors were encountered:
Is your feature request related to a problem? Please describe.
Causal graphs and their implementation inherently affect all upstream causal inference procedures. Most importantly
dodiscover
relies on the API of pywhy-graphs. Any changes introduced here should not break the "next version of dodiscover". In order to catch these, we need integration tests.Describe the solution you'd like
There are three options:
I am leaning towards adding this to
dodiscover
only, or both. We can delete one of them down the line, but the CI times are already pretty fast, so we can sacrifice this for the sake of ensuring integrability across the two packages (and eventually dowhy).The solution would just be to create a tight "accuracy" constraint on causal discovery algorithms that leverage each graph class and then this performance should not change across any changes done in
pywhy_graphs
.The text was updated successfully, but these errors were encountered: