def get_pred(x0, X_train0, n=20):
xs = [
np.insert(x[lenOfFre:lenOfFre + 21], 0,
dis(x0 - x0[10], x[0:lenOfFre] - x[10])) for x in X_train0
]
xs1 = np.asarray(xs)
#dis2= (dis(xtrain[1,0:lenOfFre],x[0:lenOfFre]) for x in xtrain[0:5,:])
ind = np.lexsort((xs1[:, 1], xs1[:, 0]))
xs_sort = xs1[ind]
xtrain_1 = xs_sort[0:n]
#xtrain_1 = xs1
#s_weight = [1/x for x in xs1[0,:]]
xx = xtrain_1[:, 1:17]
# clf = linear_model.LinearRegression()
xp5 = np.zeros(5)
for i in range(1, 6):
yy = xtrain_1[:, 16 + i]
#clf = svm.SVR(C=10000)
clf = linear_model.LinearRegression()
clf.fit(xx, yy)
xp0 = X_valid0[nx, lenOfFre:lenOfFre + 16]
xp = clf.predict(xp0)
xp5[i - 1] = xp
#print xp ,X_valid0[nx,lenOfFre+15+i]
return xp5
def check_link(v, s, threshold):
print('v shape: ', v.shape)
print('s shape: ', s.shape)
res = dis(v, s) < threshold
if res.all():
return True
else:
return False
def get_treshold(vecs):
mu = float(input('Enter the threshold hyperparameter mu: '))
sample = vecs[np.random.choice(vecs.shape[0], 500, replace=False), :]
distances = []
len = sample.shape[1]
for i in range(len):
for j in range(len):
distances.append(dis(sample[i], sample[j]))
res = np.mean(array(distances))
print('The average distance of sample: ', res)
res /= mu
return res
def check_link(v, s, threshold):
res = dis(v, s) < np.sqrt(threshold)
if res.all():
return True
else:
return False
def check_link_by_distance(v, s, threshold):
res = dis(v, s) < threshold
if res.all():
return True
else:
return False