def fraudar(self):
'''
plt.imshow(self.original_I,interpolation='none',cmap='Greys')
plt.colorbar()
plt.show()
'''
pred = list()
sparse_I = csr_matrix(self.original_I)
reviews = [(i, j) for i, j in zip(*sparse_I.nonzero())]
graph = fraudar.ReviewGraph(1, fraudar.aveDegree)
# Create reviewers and products.
reviewers = [
graph.new_reviewer("{0}".format(i)) for i in range(self.n)
]
products = [graph.new_product("{0}".format(i)) for i in range(self.m)]
# Add reviews.
for review in reviews:
graph.add_review(reviewers[review[0]], products[review[1]], 1)
# Run the algorithm.
graph.update()
# write anomalous reviewrs to output.txt
for r in graph.reviewers:
if r.anomalous_score == 1:
pred.append(r.name)
_eval = self.evaluate_fraudar(pred)
print('Fraudar: ', _eval)
return pred, _eval
def analyze(blocks):
"""Analyze a synthetic data set with a given number of blocks.
Args:
blocks: the number of blocks.
"""
graph = fraudar.ReviewGraph(blocks)
synthetic.load(graph)
graph.update()
detected = [r for r in graph.reviewers if r.anomalous_score]
correct = [r for r in detected if "anomaly" in r.name]
print(
len(correct) / len(detected),
len(correct) / synthetic.ANOMALOUS_REVIEWER_SIZE)
def create_fraudar_graph(nblock=1):
"""Create a review graph defined in Fraud Eagle package.
"""
return fraudar.ReviewGraph(int(nblock))