def parakeet_update_centroids(X, assignments, k):
d = X.shape[1]
new_centroids = np.zeros((k,d), dtype=X.dtype)
for i in xrange(k):
mask = (assignments == i)
count = np.sum(mask)
assigned_data = X[mask]
if count == 1:
new_centroids[i, :] = assigned_data[0]
elif count > 1:
new_centroids[i,:] = parakeet.mean(assigned_data)
return new_centroids
def parakeet_update_centroids(X, assignments, k):
d = X.shape[1]
new_centroids = np.zeros((k,d), dtype=X.dtype)
for i in xrange(k):
mask = (assignments == i)
count = np.sum(mask)
assigned_data = X[mask]
if count == 1:
new_centroids[i, :] = assigned_data[0]
elif count > 1:
new_centroids[i,:] = parakeet.mean(assigned_data)
return new_centroids
def test_mean(): x = np.random.randn(1000) true_mean = np.mean(x) parakeet_mean = par.mean(x) print "Numpy mean = %s, Parakeet mean = %s" % (true_mean, parakeet_mean) assert abs(true_mean - parakeet_mean) < 0.00001
def calc_centroid(X, a, i):
return par.mean(X[a == i], axis=0)