CODEWITHSUNDEEP
pythonpandasnetworkx
PV8
PV8

Reputation: 6270

Adding weights to edges in networkx automatically depending of the number of connections form pandas dataframe

I am trying to create out of a pandas dataframe a directed graph right now with networkx, so far i can use:

nx.from_pandas_edgelist(df, 'Activity', 'Activity followed', create_using=nx.DiGraph())

which shows me all the nodes and edges from Activity --> Activity followed.

In my dataframe there is sometimes the same activity followed by the same activity and i want to count this number in form of weights for the edges so far example this is my dataframe:

Index Activity Activityfollowed
0     Lunch    Dinner
1     Lunch    Dinner
2     Breakfast Lunch

should have the Edges:

Lunch --> Dinner (weight 2)

Breakfast --> Lunch (weight 1)

Is there any way to do it?

Upvotes: 3

Views: 3474

Answers (1)

Chris Adams
Chris Adams

Reputation: 18647

You could try adding the weight attribute as a column, using groupby.transform, then pass the edge_attr argument to the from_pandas_edgelist method:

df['weight'] = df.groupby(['Activity', 'Activityfollowed'])['Activity'].transform('size')

G = nx.from_pandas_edgelist(df, 'Activity', 'Activityfollowed',
                            create_using=nx.DiGraph(), edge_attr='weight')

Confirm that it's worked using:

G.edges(data=True)

[out]

OutEdgeDataView([('Lunch', 'Dinner', {'weight': 2}), ('Breakfast', 'Lunch', {'weight': 1})])

Upvotes: 6

Related Questions