word_relation.py

import training
from gensim.models import Word2Vec
import numpy as np
from sklearn.linear_model import LogisticRegressionCV
np.random.seed(0)

model = Word2Vec.load("/path/to/your/w2v_model")

data = []
X = []
y = []
for w1, w2 in training.data:
    if w1 in model and w2 in model:
      data.append([w1,w2])
      data.append([w2,w1])
    	X.append(model[w1]-model[w2])
    	X.append(model[w2]-model[w1])
    	y.append(1)
    	y.append(-1)

X = np.array(X)
y = np.array(y)

idx = np.random.permutation(len(y))
ntr = int(len(y)*0.7)
itr = idx[:ntr]
ite = idx[ntr:]

Xtr = X[itr]
ytr = y[itr]
Xte = X[ite]
yte = y[ite]

clf=LogisticRegressionCV().fit(Xtr, ytr)

ypred = clf.predict(Xte)

from sklearn import metrics
print(metrics.accuracy_score(yte, ypred)) # >>> 0.99502487562189057