def cluster_cost(Data, k, medoids, cluster_assignments):
price = np.zeros(k)
for i in range(0, k):
x = Data[cluster_assignments == i, :]
for j in range(0, np.size(x, 0)):
cost = kh.euclid_distance(x[j, :], medoids[i, :])
price[i] += cost
return price
def cluster(k, Data, medoids):
cluster_assignments = np.empty(np.size(Data, 0))
for i in range(np.size(Data, 0)):
dist = np.empty(k)
for j in range(k):
x1 = Data[i, :]
x2 = medoids[j]
dist[j] = kh.euclid_distance(x1, x2)
cluster_assignments[i] = dist.argmin()
return cluster_assignments
def find_greatest_dist(cluster_assignments, Data, medoids):
combined_array = np.empty(shape=(Data.shape[0], 2), dtype=object)
greatest_dist = np.zeros(shape=(k, 2), dtype=float)
dist = np.empty(shape=Data.shape[0])
for i in range(len(combined_array)):
combined_array[i] = [cluster_assignments[i], Data[i, :]]
x1 = Data[i, :]
currentMedoid = int(combined_array[i, 0])
x2 = medoids[currentMedoid]
dist[i] = kh.euclid_distance(x1, x2)
if greatest_dist[currentMedoid, 0] <= dist[i]:
greatest_dist[currentMedoid] = [float(dist[i]), int(i)]
# greatest_dist[:, 0] is the distance and greatest_dist[:, 1] is the id where it came from
return greatest_dist