I'm doing a machine learning project related to link prediction.But I'm stuck at reading data with networkX:
The training data I'm trying to read is stored in a "train.txt" file with the following structure:
1 2
2 3
4 3 5 1
Each line represents a node and its neighbors, ie in line 3: node 4 is connected with nodes 3, 5 and 1.
The code I'm using to read the network data is :
G = nx.read_edgelist('train.txt',delimiter = "\t",create_using = nx.DiGraph(),nodetype = int)
But this code raises a TypeError exception: failed to convert edge data as follows:
TypeError: Failed to convert edge data (['3105725', '2828522', '4394015', '2367409', '2397416',...,'759864']) to dictionary.
Welcome to SO!
Your comment is correct - this is not an edge list in the classical sense. An edge list for networkx looks something like:
1 2
2 3
4 1
4 3
4 5
Here is one way to solve your problem: read in the file line by line, and add each edge to your graph as you go.
import networkx as nx
D= nx.DiGraph()
with open('train.txt','r') as f:
for line in f:
line=line.split('\t')#split the line up into a list - the first entry will be the node, the others his friends
if len(line)==1:#in case the node has no friends, we should still add him to the network
if line[0] not in D:
nx.add_node(line[0])
else:#in case the node has friends, loop over all the entries in the list
focal_node = line[0]#pick your node
for friend in line[1:]:#loop over the friends
D.add_edge(focal_node,friend)#add each edge to the graph
nx.draw_networkx(D) #for fun draw your network
nx.read_edgelist
expects a line per edge with arbitrary data, in addition to the source and destination of the edge, so it's not what you should use in you case.
networkx offers a way to read an adjacency list from a file by using nx.read_adjlist .
Consider a file graph_adjlist.txt
.
1 2 3 4
2 5
3 5
4 5
The graph can be created according to the adjacency list as follows.
import networkx as nx
G = nx.read_adjlist('graph_adjlist.txt', create_using = nx.DiGraph(), nodetype = int)
print(G.nodes(),G.edges())
# [1, 2, 3, 4, 5] [(1, 2), (1, 3), (1, 4), (2, 5), (3, 5), (4, 5)]
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.