def test_random_events_with_taboo(self):
self.adjust_params()
events, taboos = random_events(taboo_topics=range(9),
accumulate_taboo=False,
**self.params)
for e in events:
for i in e.nodes_iter():
assert_equal(9, e.node[i]['topics'].argmax())
assert_equal(set(range(9)), taboos)
def test_random_events_with_taboo_and_accumulation(self):
self.adjust_params()
events, taboos = random_events(
taboo_topics=range(5),
accumulate_taboo=True,
**self.params
)
new_topics = set([e.node[n]['topics'].argmax()
for e in events
for n in e.nodes_iter()])
assert_equal(set(range(5, 10)), set(new_topics))
assert_equal(set(range(10)), taboos)
def test_random_events(self):
self.adjust_params()
events, _ = random_events(**self.params)
assert_equal(self.params['n_events'], len(events))
sizes = np.array([e.number_of_nodes() for e in events])
np.testing.assert_almost_equal(
self.params['event_size_mu'],
np.mean(sizes)
)
times = lambda e: [e.node[n]['timestamp'] for n in e.nodes_iter()]
mean_duration = np.mean([max(times(e)) - min(times(e))
for e in events])
np.testing.assert_almost_equal(
510,
mean_duration,
decimal=-1
)
unique_participants = lambda e: set(itertools.chain(
*[[e.node[n]['sender_id']] + e.node[n]['recipient_ids']
for n in e.nodes_iter()]
))
mean_n_participants = np.mean(
[len(unique_participants(e))
for e in events]
)
np.testing.assert_almost_equal(
5,
mean_n_participants,
decimal=0
)
for e in events:
topic_mean = np.mean([e.node[n]['topics']
for n in e.nodes_iter()],
axis=0)
topic_2nd, topic_1st = np.sort(topic_mean)[-2:]
assert_true((topic_1st / topic_2nd) > 10000)