def estimate_p_values_inter_node(net, n_samples=1000, seed=271828):
random.seed(seed)
labels = list(mmln.get_all_labels(net))
p = {}
for label in labels:
p[label] = {}
for i in range(0, len(labels)):
label1 = labels[i]
for j in range(i, len(labels)):
label2 = labels[j]
obsv = mmln.count_adjacent_labels(net, label1, label2)
n_label1 = mmln.count_labels(net, label1)
if label1 == label2:
n_label2 = n_label1
else:
n_label2 = mmln.count_labels(net, label2)
# Results at least as high as obsv
pos = 0
# Results not as high as obsv
neg = 0
for sample in range(0, n_samples):
labelled1 = random.sample(net.nodes(), n_label1)
if label1 == label2:
labelled2 = labelled1
else:
labelled2 = random.sample(net.nodes(), n_label2)
adj_labels = 0
for node1 in labelled1:
node1_neighbors = set(net.neighbors(node1))
for node2 in labelled2:
if node2 in node1_neighbors:
adj_labels += 1
if label1 == label2:
adj_labels /= 2
if adj_labels >= obsv:
pos += 1
else:
neg += 1
p[label1][label2] = pos / (pos + neg)
p[label2][label1] = p[label1][label2]
return p
def __init__(self, network):
self.n = network
self.all_labels = mmln.get_all_labels(network)
self.logger = logging.getLogger(__name__)
def test_get_all_labels(self):
net = self._get_network()
all_labels = mmln.get_all_labels(net)
self.assertEqual(len(all_labels), 2)
