将边缘列表数据转换为networkx对象

Question

I have a file which looks like-

1 2 1
1 3 1
2 999 1
2 1029 1
2 1031 1
2 1032 1
2 1197 1
2 1226 1
2 1296 1
3 450 1
3 933 1
3 934 1
3 955 1
3 1032 1
4 5 1

and I want to convert it to networkx graph but I am getting the following error-

   G=nx.read_edgelist(fh)
  File "<decorator-gen-400>", line 2, in read_edgelist
  File "E:\anaconda\lib\site-packages\networkx\utils\decorators.py", line 227, in _open_file
    result = func_to_be_decorated(*new_args, **kwargs)
  File "E:\anaconda\lib\site-packages\networkx\readwrite\edgelist.py", line 378, in read_edgelist
    data=data)
  File "E:\anaconda\lib\site-packages\networkx\readwrite\edgelist.py", line 288, in parse_edgelist
    "Failed to convert edge data (%s) to dictionary." % (d))
TypeError: Failed to convert edge data (['1']) to dictionary.

and here is the code-

fh=open("YST_full.net", 'rb')
G=nx.read_edgelist(fh)
fh.close()

What am I doing wrong here?

edit-I tried converting it to pandas dataframe

df=pd.read_csv("YST_full.net",sep=" ",names=['node1','node2','weight'])
print(df)

G=nx.from_pandas_edgelist(df, 'node1', 'node2', ['weight'])

and now I want to convert it to graphml format-

nx.write_graphml(G, "YST_full.graphml")

but the error is-

    nx.write_graphml(G, "YST_full.graphml")
  File "<decorator-gen-440>", line 2, in write_graphml_lxml
  File "E:\anaconda\lib\site-packages\networkx\utils\decorators.py", line 227, in _open_file
    result = func_to_be_decorated(*new_args, **kwargs)
  File "E:\anaconda\lib\site-packages\networkx\readwrite\graphml.py", line 149, in write_graphml_lxml
    infer_numeric_types=infer_numeric_types)
  File "E:\anaconda\lib\site-packages\networkx\readwrite\graphml.py", line 596, in __init__
    self.add_graph_element(graph)
  File "E:\anaconda\lib\site-packages\networkx\readwrite\graphml.py", line 658, in add_graph_element
    T = self.xml_type[self.attr_type(k, "edge", v)]
KeyError: <class 'numpy.int64'>

Answer 1

You must inform networkx that the third column is an attribute called "weight" (or whatever you call it):

graph = nx.read_edgelist("YST_full.net", data=(('weight', float),))

As far as your second question is concerned, sometimes networkx fails to convert a NumPy int64 to the Python int before exporting to GraphML. You have to do it yourself:

weights = {(n1,n2): float(d['weight']) # or int()
                    for n1,n2,d in graph.edges(data=True)}
nx.set_edge_attributes(G, weights, 'weight')

Answer 2

This error is due to the dtypes of your pandas dataframe. A workaround is to convert your dataframe columns to string dtype.

df = df.apply(lambda x: x.astype(str))
G=nx.from_pandas_edgelist(df, 'node1', 'node2', 'weight')
nx.write_graphml(G,'test.out')

Output:

<?xml version='1.0' encoding='utf-8'?>
<graphml xmlns="http://graphml.graphdrawing.org/xmlns" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://graphml.graphdrawing.org/xmlns http://graphml.graphdrawing.org/xmlns/1.0/graphml.xsd"><key attr.name="weight" attr.type="string" for="edge" id="d0"/>
<graph edgedefault="undirected"><node id="1"/>
<node id="2"/>
<node id="3"/>
<node id="999"/>
<node id="1029"/>
<node id="1031"/>
<node id="1032"/>
<node id="1197"/>
<node id="1226"/>
<node id="1296"/>
<node id="450"/>
<node id="933"/>
<node id="934"/>
<node id="955"/>
<node id="4"/>
<node id="5"/>
<edge source="1" target="2">
  <data key="d0">1</data>
</edge>
<edge source="1" target="3">
  <data key="d0">1</data>
</edge>
<edge source="2" target="999">
  <data key="d0">1</data>
</edge>
<edge source="2" target="1029">
  <data key="d0">1</data>
</edge>
<edge source="2" target="1031">
  <data key="d0">1</data>
</edge>
<edge source="2" target="1032">
  <data key="d0">1</data>
</edge>
<edge source="2" target="1197">
  <data key="d0">1</data>
</edge>
<edge source="2" target="1226">
  <data key="d0">1</data>
</edge>
<edge source="2" target="1296">
  <data key="d0">1</data>
</edge>
<edge source="3" target="450">
  <data key="d0">1</data>
</edge>
<edge source="3" target="933">
  <data key="d0">1</data>
</edge>
<edge source="3" target="934">
  <data key="d0">1</data>
</edge>
<edge source="3" target="955">
  <data key="d0">1</data>
</edge>
<edge source="3" target="1032">
  <data key="d0">1</data>
</edge>
<edge source="4" target="5">
  <data key="d0">1</data>
</edge>
</graph></graphml>

Answer 3

Today I started learning network analysis and this is the first error I encountered. I just changed G=nx.read_edgelist(fh) to G=nx.read_weighted_edgelist(fh) .

You can also remove the third column and use G=nx.read_edgelist(fh)

Answer 4

If you do change G=nx.read_edgelist(fh) to G=nx.read_weighted_edgelist(fh) , it will work.

You don't need to remove third column. I am using SNAP Temporal data for node2vec project, was facing same error as described, solving by using G = nx.read_weighted_edgelist(args.input, nodetype=int, data=(('weight',float),),delimiter=',', create_using=nx.DiGraph())