CODEWITHSUNDEEP
pythongraphnetworkx
constantLeaner
constantLeaner

Reputation: 167

networkx: giving a wrong direction for an edge

I am trying to read some dataframe to a networkx graph like this:

bipartGraph = nx.Graph()
bipartGraph.add_edge(618254814, 14337833)
bipartGraph.add_edge(618254882, 12087274)

When I display the edges using the bipartGraph.edges() function, I get the below:

[(14337833, 618254814), (618254882, 12087274)]

So, the direction of the first edge is reversed. I am trying to build a bipartite graph from a dataframe which I need to reuse to build an another graph. Is there any specific property of networkx I am missing?

Upvotes: 0

Views: 1167

Answers (1)

Joel
Joel

Reputation: 23887

At least in networkx 1.11 if your graph is undirected, an edge that is added as (u,v) may be returned as either (v,u) or (u,v). [I think this may have been changed in version 2.0]. This is because the underlying data structure is a dict, and there is no guarantee that a dict returns values in any particular order: Why items order in a dictionary changed in Python?.

If you really want a direction on your edge, you should make your graph a DiGraph.

Upvotes: 4

Related Questions