def run_iris(X, y, l, name, plot_lambda=False):
#iris = datasets.load_iris()
#X = iris.data
#y = iris.target
#X, y = make_classification(1000, 10)
lambda_interval = np.linspace(0., 1., 11)
time1 = time.clock()
AWC_object = AWC(speed=1.)
# To tune parameter \lambda, plot sum of the weights for \lambda 's from some interval
#and take a value at the end of plateau or before huge jump.
if plot_lambda == True:
AWC_object.plot_sum_of_weights(lambda_interval, X)
if l is None:
l = 0.6
AWC_object.awc(l, X)
clusters = AWC_object.get_clusters()
labels = AWC_object.get_labels()
time2 = time.clock()
print(time2 - time1)
draw(X, labels, name, True)
print('Estimated number of clusters: %d' % len(set(labels)))
print('cluster sizes: '),
for c in clusters:
print(len(c)),
print("\nV-measure: %0.3f" % metrics.v_measure_score(y, labels))
class MixedXor:
def __init__(self, j, k, m, initial, j2, k2, m2, initial2):
self.first = AWC(j, k, m, initial)
self.second = FibAdd(j2, k2, m2, initial2)
def get_next(self):
return self.first.get_next() ^ self.second.get_next()
def __init__(self, j, k, m, initial, j2, k2, m2, initial2):
self.first = AWC(j, k, m, initial)
self.second = FibAdd(j2, k2, m2, initial2)
