Error about separate graphs when learning BN

[sersoage]

Description

I am getting the following error "The given structure has 5 separated graph components. Please make sure it has only one."
Structure was learnt with notears but when i run bn = BayesianNetwork(sm) i get the above error. Any advice? How can I select or display the particular subgraphs

Context

Trying to compute CPT and display markov blanket

[oentaryorj]

Thanks for the question.

Ideally, the structure model to be fed into BayesianNetwork should be a single connected component with no isolated nodes.

You can use plot_structure to inspect the graph produced by NOTEARS, as per this tutorial, and then manually remove nodes or edges based on, e.g., input from domain experts. We also have a helper function sm.get_largest_subgraph() to extract the largest connected component in the graph, as well as sm.get_target_subgraph() to extract the subgraph for a given node.

Alternatively, to iterate over all subgraphs, you can use the function weakly_connected_components in networkx. For example:

import networkx as nx

for sub_graph in nx.weakly_connected_components(sm):
     print(list(sub_graph.edges()))

Do let us know if the above helps.

[oentaryorj]

Closing this for now. Feel free to open a new issue if you have another enquiry. Thank you.