def test_with_hartigan(self):
"""A tiny dataset which can't possibly work here"""
dataset = testloader.load_hartigan()
with self.assertRaises(ValueError):
bfinit.generate(dataset.data, 3)
def run_clustering(algorithm):
"""Run all datasets"""
_run_dataset('HART', testloader.load_hartigan(), 3, algorithm)
_run_dataset('IRIS', testloader.load_iris(), 3, algorithm)
_run_dataset('SOYS', testloader.load_soy_small(), 4, algorithm)
_run_dataset('WINE', testloader.load_wine(), 3, algorithm)
_run_dataset('WBCO', testloader.load_wbco(), 2, algorithm)
def _test_with_hartigan(self):
"""This is left as TODO as it exposes the fact that the algorithm is
highly sensitive to the values of R and C being specified for each
dataset. This seems potentially problematic"""
dataset = testloader.load_hartigan()
centroids = macqueen.generate(dataset.data, 3)
self.assertEqual((3, 3), centroids.shape)
import sys
from sklearn.cluster import KMeans
from datasets import testloader
from initialisations import spss2010 as alg
dsname = ''
if len(sys.argv) > 1:
dsname = sys.argv[1]
if dsname == 'iris':
dataset = testloader.load_iris()
num_clusters = 3
else:
dataset = testloader.load_hartigan()
num_clusters = 3
# dataset = testloader.load_soy_small()
# num_clusters = 4
data = dataset.data
centroids = alg.generate(data, num_clusters)
print(centroids)
est = KMeans(n_clusters=num_clusters, init=centroids, n_init=1)
est.fit(dataset.data)
print("Final centres:\n", est.cluster_centers_)
def test_with_hartigan(self):
"""A tiny dataset which can't possibly work here""" # why not, Simon?
dataset = testloader.load_hartigan()
centroids = ikminit_c.generate(dataset.data, 3)
self.assertEqual((3, 3), centroids.shape)
def test_with_hartigan(self):
"""A tiny dataset which led to problems with empty clusters"""
dataset = testloader.load_hartigan()
centroids = hand.generate(dataset.data, 3)
self.assertEqual((3, 3), centroids.shape)