def create_ARFF_network_metrics_file(g,
node_to_score,
seeds,
arff_file_name,
calculate_topological_values=False):
delim = ","
header = "@RELATION aneurysm\n@ATTRIBUTE id STRING\n@ATTRIBUTE score NUMERIC\n" + \
"@ATTRIBUTE degree INTEGER\n@ATTRIBUTE linker_degree INTEGER\n" + \
"@ATTRIBUTE ld_ratio NUMERIC\n@ATTRIBUTE clustering_coefficient NUMERIC\n" + \
"@ATTRIBUTE betweenness_centrality NUMERIC\n" + \
"@ATTRIBUTE degree2 INTEGER\n@ATTRIBUTE linker_degree2 INTEGER\n" + \
"@ATTRIBUTE ld_ratio2 NUMERIC\n" + \
"@ATTRIBUTE class {involved,not-involved}\n@DATA\n"
seeds = set(seeds)
if calculate_topological_values:
print "Calculating betweenness centrality.."
mapB = networkx.betweenness_centrality(g)
##mapB = dict(zip(g.nodes(), range(len(g.nodes()))))
if calculate_topological_values:
print "Calculating clustering coefficients.."
mapC = networkx.clustering(g, with_labels=True)
#print "connected component sizes: ", map(len, networkx.connected_components(g))
#cliques = networkx.find_cliques(g) # high computational cost
node_to_values = get_node_degree_related_values(g, seeds)
f = open(arff_file_name, 'w')
f.write(header)
for v in g.nodes_iter():
d, ld, d2, ld2 = node_to_values[v]
if calculate_topological_values:
cc = mapC[v]
bc = mapB[v]
else:
cc = 0.0
bc = 0.0
if d == 0: r1 = 0.0
else: r1 = float(ld) / d
if d2 == 0: r2 = 0.0
else: r2 = float(ld2) / d2
if v in seeds:
if node_to_score is not None:
s = node_to_score[v]
else:
s = 1
c = "involved"
else:
s = "?"
c = "not-involved"
# id score degree linker_degree ld_ratio clustering_coeff betweenness_cent d2 ld2 ld_ratio2 class
# v s d ld n1 cc bc d2 ld2 n2 c
f.write(("%s" % delim
).join(map(str, [v, s, d, ld, r1, cc, bc, d2, ld2, r2, c])) +
"\n")
f.close()
return
def create_ARFF_network_metrics_file(g, node_to_score, seeds, arff_file_name, calculate_topological_values = False):
delim = ","
header = "@RELATION aneurysm\n@ATTRIBUTE id STRING\n@ATTRIBUTE score NUMERIC\n" + \
"@ATTRIBUTE degree INTEGER\n@ATTRIBUTE linker_degree INTEGER\n" + \
"@ATTRIBUTE ld_ratio NUMERIC\n@ATTRIBUTE clustering_coefficient NUMERIC\n" + \
"@ATTRIBUTE betweenness_centrality NUMERIC\n" + \
"@ATTRIBUTE degree2 INTEGER\n@ATTRIBUTE linker_degree2 INTEGER\n" + \
"@ATTRIBUTE ld_ratio2 NUMERIC\n" + \
"@ATTRIBUTE class {involved,not-involved}\n@DATA\n"
seeds = set(seeds)
if calculate_topological_values:
print "Calculating betweenness centrality.."
mapB = networkx.betweenness_centrality(g)
##mapB = dict(zip(g.nodes(), range(len(g.nodes()))))
if calculate_topological_values:
print "Calculating clustering coefficients.."
mapC = networkx.clustering(g, with_labels=True)
#print "connected component sizes: ", map(len, networkx.connected_components(g))
#cliques = networkx.find_cliques(g) # high computational cost
node_to_values = get_node_degree_related_values(g, seeds)
f = open(arff_file_name, 'w')
f.write(header)
for v in g.nodes_iter():
d, ld, d2, ld2 = node_to_values[v]
if calculate_topological_values:
cc = mapC[v]
bc = mapB[v]
else:
cc = 0.0
bc = 0.0
if d == 0: r1 = 0.0
else: r1 = float(ld)/d
if d2 == 0: r2 = 0.0
else: r2 = float(ld2)/d2
if v in seeds:
if node_to_score is not None:
s=node_to_score[v]
else:
s = 1
c="involved"
else:
s="?"
c="not-involved"
# id score degree linker_degree ld_ratio clustering_coeff betweenness_cent d2 ld2 ld_ratio2 class
# v s d ld n1 cc bc d2 ld2 n2 c
f.write( ("%s" % delim).join( map(str, [v, s, d, ld, r1, cc, bc, d2, ld2, r2, c]) ) + "\n" )
f.close()
return
def get_clustering_coefficient_map(g):
return networkx.clustering(g, with_labels=True)
def get_clustering_coefficient_map(g):
return networkx.clustering(g, with_labels=True)