def naiveRecommendFromCluster(x, cluster):
N, D = cluster.shape
assert(x.shape == (D,))
m = 0
min_diff = 100000
for i in range(N):
diff = ml.euclideanDistance(x, cluster[i])
if diff < min_diff:
min_diff = diff
m = i
a = recommendFromElement(x, cluster[m])
return a
import config
amount=config.amount
ranks, website_names, X = data.getBandHistograms(amount=amount, cut=config.cut, big=config.big)
N, D = X.shape
print "Each feature vector has dimension %d" % D
print "Training on %d samples" % N
# Similarities matrix
S = np.zeros((N, N))
for i in xrange(N):
for j in xrange(N):
S[i, j] = -ml.euclideanDistance(X[i], X[j])
# Preference i p(i) = S(i,i)
# This is the a priori suitability of point i to serve as an exemplar
# We initialize all of them to the median
median = np.median(S)
for i in xrange(N):
S[i, i] = median
print S
# Responsabilities matrix
R = np.zeros((N, N))
# Availability matrix
A = np.zeros((N, N))
C = np.array(xrange(N))
