CODEWITHSUNDEEP
pythonpandasnetworkx
David
David

Reputation: 89

Pandas DataFrames to Graph

Given two DataFrames, one for Nodes and another for Edges. How can I get those into a networkx graph including the attribute columns I have?

Example:

df_nodes

ID Label Attribute_W Attribute_X
0 0 Japan Asia 81
1 1 Mexico America 52
2 2 Ireland Europe 353

df_Edges

Target Source Attribute_Y Attribute_Z
0 0 1 10 1
1 0 2 15 2
2 1 2 20 3

I tried with G = nx.from_pandas_edgelist but it returns an attribute error. And I'm not sure of how to add the attributes in the construction of the graph.

I'm looking to output a graphml file nx.write_graphml(G, "My_File.graphml")

Thanks.

Upvotes: 1

Views: 455

Answers (1)

kappablanca
kappablanca

Reputation: 76

I too do not see any way how to include node attributes via from_pandas_edgelist. However, you can achieve what you want just using nx.add_edges_from and nx.add_nodes_from in a few lines of code.

G = nx.Graph()

node_label_attr = "Label"
for index,row in df_nodes.iterrows():
    as_dict = row.to_dict()
    label = as_dict[node_label_attr]
    del as_dict[node_label_attr]
    G.add_nodes_from([(label, as_dict)])

for index,row in df_Edges.iterrows():
    as_dict = row.to_dict()
    source = as_dict["Source"]
    target = as_dict["Target"]
    del as_dict["Source"]
    del as_dict["Target"]
    G.add_edges_from([(source,target,as_dict)])

You iterate through the rows of the dataframe and convert them to dictionaries which are understood by nx.add_nodes_from and nx.add_edges_from.

Upvotes: 2

Related Questions