QUESTION: To get DAG/Graph from FCM-based methods

[ukamath]

With FCM-based methods, for instance DirectiINGAM, how do we get DAG so that we can compare and contrast the structure with other methods such as PC,FCI, GES etc? Are there any utilities?

from causallearn.search.FCMBased import lingam model = lingam.DirectLiNGAM() model.fit(data)

print(model.causalorder) print(model.adjacencymatrix)

Thanks

[ukamath]

Got it working using DiGraph

import networkx as nx import matplotlib.pyplot as plt

create DAG using NetworkX

dag = nx.DiGraph(model.adjacencymatrix)

relabel nodes with variables X1, X2, ..., XN

node_labels = {i: f"X{i+1}" for i in range(len(dag.nodes))} dag = nx.relabel_nodes(dag, node_labels)

draw DAG using Matplotlib

nx.draw(dag, with_labels=True) plt.show()