import networkx as nx
def numTriangles(G, n):
EG = nx.ego_graph(G, n)
#each triangle is counted three times, once at each node
numTri = sum(nx.triangles(EG).values()) / 3
print("number of triangles in ego-network: " + str(numTri))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
numTriangles(convert(politicalSet), 0)
import networkx as nx
def sizeConnectedComponents(G):
num = nx.number_connected_components(G)
print("number of connected components: " + str(num))
lenEachCC = [len(c) for c in nx.connected_components(G)]
print("list of their sizes: " + str(lenEachCC))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
sizeConnectedComponents(convert(emailSet))
import networkx as nx
def numEdges(G, n):
EG = nx.ego_graph(G, n)
numEdges = EG.number_of_edges()
print("number of edges in ego-network: " + str(numEdges))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
numEdges(convert(emailSet), 212)
import networkx as nx
def numTriangles(G):
#each triangle is counted three times, once at each node
num = sum(nx.triangles(G).values())/3
print("Number of triangles: " + str(num))
def numTriangles2():
G = nx.Graph()
G.add_edges_from([(1, 2), (1, 3), (2, 3), (3, 4)])
num = sum(nx.triangles(G).values())/3
print(nx.triangles(G))
print("Number of triangles: " + str(num))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
graph = convert(emailSet)
#numTriangles(graph)
numTriangles2()
G = genPPM.generate()
print(G.number_of_nodes())
print(G.number_of_edges())
n = G.number_of_nodes()
nArr = []
for i in range(n):
nArr.append(i + 1)
print(nArr)
#Express graph as graph adjacency matrix / Numpy matrix
M = nx.to_numpy_matrix(G)
#Find all singular values of matrix
s = np.linalg.svd(M)[1]
print(s)
#Draw loglog graph
plt.plot(np.log10(nArr),
np.log10(s),
linestyle='None',
marker='x',
markeredgecolor='blue')
plt.xscale('log')
plt.yscale('log')
plt.show()
from src.preprocess.csvToGraph import convert
politicalSet = 'data/datasetPolitical/datasetPolitical_final2.csv'
computeTopSVal(convert(politicalSet), 600)
kArr.append(k)
accuracyArr.append(accuracy)
k = k + 1
plotkError(kArr, accuracyArr)
def plotkError(kArr, accuracyArr):
plt.plot(kArr,
accuracyArr,
linestyle='None',
marker='x',
markeredgecolor='blue')
plt.title(
"k={1...50}, Linear-linear graph, average classification accuracy vs. k, political set"
)
plt.ylabel("Classification Accuracy")
plt.xlabel("k")
plt.show()
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
#from src.svd.computeTruncated import computeTrun
#M = computeTrun(convert(politicalSet))
#classifyData(M)
classifyDataGeneric(convert(politicalSet))
import networkx as nx
import scipy as sp
def largestSingularValue(G, n):
EG = nx.ego_graph(G, n)
#Express graph as graph adjacency matrix / SciPy sparse matrix
M = sp.sparse.csc_matrix(nx.to_scipy_sparse_matrix(EG)).asfptype()
#Find greatest singular value of matrix
s = sp.sparse.linalg.svds(M, 1)[1]
print("largest singular value of adjacency of ego-network: " + str(s))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
largestSingularValue(convert(emailSet), 212)
import networkx as nx
#import snap
def diameter(G):
diameter = nx.diameter(G)
print("diameter: " + str(diameter))
#def effectiveDiameter(G):
#effDiam = snap.GetBfsEffDiam(G, 10, false)
#print("90-percentile effective diameter: " + str(effDiam))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
diameter(convert(politicalSet))
#effectiveDiameter(graph)
indexed_anomaly_score = []
for i in range(len(anomaly_score)):
indexed_anomaly_score.append([anomaly_score[i], testIndexes[i]])
#print("indexed anomaly score: ", indexed_anomaly_score)
indexed_anomaly_score.sort(key=sortFirst)
#print("indexed anomaly score sorted: ", indexed_anomaly_score)
count = 0
for i in range(n):
if indexed_anomaly_score[i][1] in outlierIndexes:
print("found correct anomaly index: ",
str(indexed_anomaly_score[i][1]))
count = count + 1
print("count: " + str(count))
accuracy_score = float(count) / n
print("accuracy score: " + str(accuracy_score))
accuracyArr.append(accuracy_score)
averageAccuracy = sum(accuracyArr, 0.0) / len(accuracyArr)
print("average accuracy score: " + str(averageAccuracy))
return averageAccuracy
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
M = computeAllProperties(convert(politicalSet))
detectIsolation(M)
)
plt.ylabel("Anomaly Accuracy")
plt.xlabel("k")
plt.show()
def printMatrix(infile):
lookup_classify = pickle.load(
open('data/datasetPolitical/blogs_classify.pkl', 'rb'))
first = [k for k, v in lookup_classify.items() if v == 0]
second = [k for k, v in lookup_classify.items() if v == 1]
df = pd.read_csv(infile)
df.columns = ['source', 'target', 'weight']
plt.plot(df[df.columns[0]],
df[df.columns[1]],
linestyle='None',
marker='.',
markeredgecolor='blue')
plt.ylabel("col")
plt.xlabel("row")
plt.show()
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final1.csv'
detectIsolationGeneric(convert(politicalSet))
#printMatrix(politicalSet)
#Calculate Frobenius norm for matrix A-B
normAB = sp.linalg.norm(A - B)
error = normAB / normA
if (error <= 0.1):
finalU = U
kArr.append(k)
errorArr.append(error)
k = k + 20
plotkError(kArr, errorArr)
return finalU
def plotkError(kArr, errorArr):
#Draw loglog graph
plt.plot(kArr, errorArr, linestyle='None',
marker='x', markeredgecolor='blue')
plt.title("Linear-linear graph, political set")
plt.ylabel("Reconstruction Error")
plt.xlabel("k")
#plt.xscale('log')
#plt.yscale('log')
plt.show()
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
computeTrun(convert(politicalSet))
import networkx as nx
def localClustering(G, n):
c = nx.clustering(G, n)
print("local clustering coefficient: " + str(c))
from src.preprocess.csvToGraph import convert
emailSet = 'data/datasetEmail/datasetEmail_final.csv'
politicalSet = 'data/datasetPolitical/datasetPolitical_final.csv'
localClustering(convert(emailSet), 212)