def kmeans(samples, k, verbose):
"""Assumes samples is a list of samples of class Sample,
k is a positive int, verbose is a Boolean
Returns a list containing k clusters. """
# Get k randomly chosen initial centroids
initialCentroids = random.sample(samples, k)
# Create a singleton cluster for each centroid
clusters = []
for e in initialCentroids:
clusters.append(cluster.Cluster([e]))
# Iterate until centroids do not change
converged = False
numIterations = 0
while not converged:
numIterations += 1
# replace the following line by implementing
# kmeans_iter(samples, clusters, k) in this file
converged = helper.kmeans_iter(samples, clusters, k)
# converged = kmeans_iter(samples, clusters, k)
if verbose:
print('Iteration #' + str(numIterations))
for c in clusters:
print(c)
print('\n') # add blank line
return clusters
def kmeansTest(k=2, n=20, verbose=False):
random.seed(0)
xMean = 3
xSD = 1
yMean = 5
ySD = 1
d1Samples = util.genDistribution(xMean, xSD, yMean, ySD, n, '1.')
d2Samples = util.genDistribution(xMean+3, xSD, yMean+1, ySD, n, '2.')
allSamples = d1Samples + d2Samples
print("before clustering")
util.plot_cluster([cluster.Cluster(allSamples)])
print("after clustering")
clusters = kmeans(allSamples, k, verbose)
util.plot_cluster(clusters, verbose)
print('Final result')
for c in clusters:
print('', c)
# 's', 'p', '*', 'h', 'H', 'D', 'd')
# colors = ('b', 'g', 'c', 'm', 'y', 'k')
# return [c + m for m in markers for c in colors]
def plot_cluster(clusters, verbose = False, centroid = True):
MARKERS = make_cmarkers()
COLORS = make_cmap()
for l in range(len(clusters)):
c = clusters[l]
cm = COLORS[l]+ MARKERS[l]
plotSamples(c.getMembers(), cm, verbose)
if centroid:
plotSamples([c.centroid], 'sr')
plt.show()
if __name__ == "__main__":
#print(minkowskiDist([0, 0], [1, 1], 1))
#print(minkowskiDist([0, 0], [1, 1], 2))
test_samples = genDistribution()
c = cluster.Cluster(test_samples)
plot_cluster([c])
# plotSamples(test_samples, 'o')
# plotSamples([test_samples[0]], 'sk')
# plt.show()
# make data
random.seed(0)
n = 100
K = 3
LABELS = ('a', 'b', 'c')
all_cluster = []
data = []
for i in range(K):
tmp_data = util.genDistribution(i * 2 + 1,
1,
i * 2 + 1,
1,
n=20,
label=LABELS[i])
all_cluster.append(cl.Cluster(tmp_data))
data += tmp_data
def onclick(event):
# Creating a new point and finding the k nearest neighbours
new = sample.Sample('', [event.xdata, event.ydata], '')
knn(new, data, K)
# draw the new point
data.append(new)
pylab.scatter([new.getFeatures()[0]], \
[new.getFeatures()[1]], \
label = new.getLabel(), \
marker = util.make_cmarkers()[LABELS.index(new.getLabel())], \
color = util.make_cmap()[LABELS.index(new.getLabel())])
pylab.draw()
