def find_cluster(seg_snt, means):
max_score = -10000.0
m_incex = -1
for i in range(len(means)):
score = sim.cos(seg_snt.vector, means[i])
if score > max_score:
max_score = score
m_incex = i
return m_incex, max_score
def calEntityPairScoreTransE(self, enIdx1, enIdx2):
'''
Return TransE score:
[('father': 0.9), ('mother': 0.5), ...]
'''
e1 = self.train.get_all_entity()[enIdx1]
e2 = self.train.get_all_entity()[enIdx2]
all_relation = self.train.get_all_relation()
# scores = [(all_relation[i], cos(self.E[all_relation[i]], minusVector(self.E[e2], self.E[e1]))) for i in range(len(self.R))]
scores = [cos(self.E[all_relation[i]], minusVector(self.E[e2], self.E[e1])) for i in range(len(self.R))]
return scores
def calEntityPairScoreTransE(self, enIdx1, enIdx2):
'''
Return TransE score:
[('father': 0.9), ('mother': 0.5), ...]
'''
try:
e1 = self.train.get_all_entity()[enIdx1]
e2 = self.train.get_all_entity()[enIdx2]
all_relation = self.train.get_all_relation()
# print e1, e2, len(self.R), len(self.Rel)
# scores = [(all_relation[i], cos(self.E[all_relation[i]], minusVector(self.E[e2], self.E[e1]))) for i in range(len(self.R))]
scores = [cos(self.Rel[all_relation[i]], minusVector(self.Ent[e2], self.Ent[e1])) for i in range(len(self.R))]
except Exception as e:
print e
print all_relation[e1], all_relation[e2]
scores = []
return scores
def calc_doc_sim(sdoci, sdocj):
return sim.cos(sdoci.vector, sdocj.vector)
print('B:', sim.support_mult(p_list, set2))
print('C:', sim.support_mult(p_list, set3))
print('D:', sim.support_mult(p_list, set4))
# naive bayes
class_list = [0, 0, 0, 1, 1, 1, 2, 2, 2]
indx_list = [0, 1]
equal_to = [1, 0]
p = sim.naive_bayes(p_list, class_list, 2, indx_list, equal_to, 3)
print('p:', p)
# similarity measures
smc = sim.smc(P2, P6)
j = sim.j_coeff(P2, P7)
cos6 = sim.cos(P2, P6)
cos7 = sim.cos(P2, P7)
print('A:', smc > j)
print('B:', smc > cos6)
print('C:', cos7 > j)
print('D:', cos6 > cos7)
# confidence
# X --> Y
# a_left = X, a_right = Y
a_left = [1, 3, 5]
a_right = [0, 4]
# lift, no script for this but very easy to calculate
# with the given conf and support script
a_right = [hpH]
print('conf:', sim.conf(o_list, a_left, a_right))
print('A:', sim.support_mult(o_list, A))
print('B:', sim.support_mult(o_list, B))
print('C:', sim.support_mult(o_list, C))
print('D:', sim.support_mult(o_list, D))
a = [1, 0, 1, 0, 0, 1]
b = [1, 0, 1, 0, 1, 0]
c = [0, 0, 0, 0, -1, 1]
print('A:', sim.euclid_norm(c))
print('B:', sim.p_norm(c, 1) < sim.euclid_norm(c))
print('C:', sim.j_coeff(a, b))
print('D:', sim.cos(a, b) == sim.smc(a, b))
# high: z = 0
# low: z = 1
class_list = [1, 1, 1, 1, 0, 0, 0, 1]
C = 2
indx_list = [hpL, am0]
class_check = 0
p = sim.naive_bayes(o_list, class_list, class_check, indx_list, [1, 1], C)
print('p:', p)
p1 = [0, 0.2606, 1.1873, 2.4946, 2.9510, 2.5682, 3.4535, 2.4698]
p2 = [0.2606, 0, 1.2796, 2.4442, 2.8878, 2.4932, 3.3895, 2.4216]
p3 = [1.1873, 1.2796, 0, 2.8294, 3.6892, 2.9147, 4.1733, 2.2386]
p4 = [2.4946, 2.4442, 2.8294, 0, 1.4852, 0.2608, 2.2941, 1.8926]
a_left = [0, 1, 2, 3, 4]
a_right = [5]
print('conf:', sim.conf(p_list, a_left, a_right))
indx_list = [0, 1]
class_list = [1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0]
equal_to = [1, 1]
p = sim.naive_bayes(p_list, class_list, 1, indx_list, equal_to, 2)
print('p:', p)
j = sim.j_coeff(P1, P3)
smc = sim.smc(P1, P3)
cos = sim.cos(P1, P3)
print('A:', j < smc)
print('B:', j > cos)
print('C:', smc > cos)
print('D:', cos == 3/15)
print('d', sim.p_norm([-0.25, -0.25], 1))
e = [True, True, True, True, True]
e1 = [False] * 20
print('ada:', sim.ada_boost([e + e1]))
root = [3, 1, 3]
splits = [[1, 1, 0], [2, 0, 3]]
print('dec:', sim.dec_tree_ce(root, splits))
o1 = [4, 7, 9, 5, 5, 5, 6]
o2 = [4, 7, 7, 7, 3, 7, 8]
o3 = [7, 7, 10, 6, 6, 4, 9]
o4 = [9, 7, 10, 8, 6, 10, 9]
o5 = [5, 7, 6, 8, 8, 6, 7]
o6 = [5, 3, 6, 6, 8, 8, 11]
o7 = [5, 7, 4, 10, 6, 8, 7]
o8 = [6, 8, 9, 9, 7, 11, 7]
o_list = [o1, o2, o3, o4, o5, o6, o7, o8]
print('A:', sim.cos(o1, o3))
print('B:', sim.j_coeff(o1, o3))
print('C:', sim.smc(o1, o3))
class_list = [1, 1, 1, 1, 0, 0, 0, 0]
print('knn:', sim.knn(o_list, class_list, 1, [0, 1]))
ard = sim.knn_density(o1, 1) / sim.knn_density(o2, 1)
print('ard:', ard)
s1 = [0, 1, 1, 0, 1, 0]
s2 = [0, 1, 1, 1, 0, 1]
s3 = [1, 1, 1, 0, 1, 0]
s4 = [1, 1, 1, 0, 1, 0]
s5 = [0, 1, 1, 0, 1, 1]
s6 = [0, 0, 1, 1, 1, 1]