def fit(self, X):
M, Xm = self._compute_m(X)
U = linalg.svds(M, k=2)[0]
normal = np.cross(U.T[0], U.T[1])
#normal = numpy.linalg.svd(M)[0][:, 2]
if normal[2] < 0:
normal *= -1
dist = np.dot(normal, Xm)
return dist, normal, M
def SVD():
user_list, user_brand_matrix = generate_userbrand_matrix()
U, Sigma, VT = la.svds(user_brand_matrix, k=71)
Sigma = np.array(Sigma)
U = np.mat(U)
VT = np.mat(VT)
'''
print Sigma
sig2 = Sigma**2
print sum(sig2)
print sum(sig2)*0.9
print sum(sig2[-10:])/sum(sig2)
print sum(sig2[-15:])/sum(sig2)
print sum(sig2[-20:])/sum(sig2)
'''
Sig20 = np.mat(np.eye(20) * Sigma[-20:])
print(user_brand_matrix * (VT[-20:, :].T) * Sig20).shape
SVD_matrix = user_brand_matrix * (VT[-20:, :].T) * Sig20.I
Dict = dict()
index = 0
for u in user_list:
Dict[u] = np.array(SVD_matrix[index]).reshape(20)
index += 1
persistence(Dict, 'SVD_dict.txt')
def SVD():
user_list,user_brand_matrix = generate_userbrand_matrix()
U,Sigma,VT = la.svds(user_brand_matrix,k=71)
Sigma = np.array(Sigma)
U = np.mat(U)
VT = np.mat(VT)
'''
print Sigma
sig2 = Sigma**2
print sum(sig2)
print sum(sig2)*0.9
print sum(sig2[-10:])/sum(sig2)
print sum(sig2[-15:])/sum(sig2)
print sum(sig2[-20:])/sum(sig2)
'''
Sig20 = np.mat(np.eye(20)*Sigma[-20:])
print (user_brand_matrix*(VT[-20:,:].T)*Sig20).shape
SVD_matrix = user_brand_matrix*(VT[-20:,:].T)*Sig20.I
Dict = dict()
index = 0
for u in user_list:
Dict[u] = np.array(SVD_matrix[index]).reshape(20)
index +=1
persistence(Dict,'SVD_dict.txt')
for row in X:
mu = np.add(mu, row)
mu = np.divide(mu, N)
centeredX = np.zeros((N, d))
j = 0
for row in X:
centeredX[j, :] = row - mu
j = j + 1
# print("centeredX",centeredX[0,:])
# print("mu",mu)
# print("X",X[0,:])
#==================================================================
# Singular Value decomposition of the centered Data
#==================================================================
p = 10
u, s, vt = sla.svds(centeredX, k=p)
Vp = vt.T
Z = np.matmul(centeredX, Vp)
#==================================================================
# Reconstruction
#==================================================================
newX = np.zeros((N, d))
newX = np.matmul(Z, Vp.T)
j = 0
for row in newX:
newX[j, :] = row + mu
j = j + 1
print(newX.shape)
print(Vp.shape)
#==================================================================
# Save images