def compute_distortion():
data_table = alp.load_data_table(alp.DATA_111_URL)
cluster_list = alp.run_example()
list1 = []
for i in cluster_list:
error = i.cluster_error(data_table)
list1.append(error)
print sum(list1)
def question10(URLs, min_number_clusters=6, max_number_clusters=20, steps=5, file_to_save="Question10.png"):
'''
:return:
'''
for url in URLs:
hierarchical_distortion_result = []
kmeans_distortion_result = []
for num in range(max_number_clusters,min_number_clusters-1,-1):
data_table, singleton_list = alg_project3_viz.run_example(url)
cluster_list_hierarchical = hierarchical_clustering(singleton_list, num)
hierarchical_distortion = sum([x.cluster_error(data_table) for x in cluster_list_hierarchical])
hierarchical_distortion_result.append(hierarchical_distortion)
hierarchical_distortion_result.reverse()
for num in range(min_number_clusters, max_number_clusters + 1):
data_table, singleton_list = alg_project3_viz.run_example(url)
#print "num clsusters", num
cluster_list_kmeans = kmeans_clustering(singleton_list, num, steps)
kmeans_distortion = sum(cluster.cluster_error(data_table) for cluster in cluster_list_kmeans)
#print "kmeans distortion", kmeans_distortion
kmeans_distortion_result.append(kmeans_distortion)
counties = re.search('_(\d+)\.csv', url).group(1)
print "hierarchical clustering with", num, "cluster for", counties, "counties", hierarchical_distortion_result
print "k-means clustering with", num, "cluster for", counties, "counties", kmeans_distortion_result, "\n"
label_hierar = 'Distortion of hierarchical clustering for' + counties
label_kmeans = 'Distortion of k-means clustering for' + counties
range0 = range(min_number_clusters, max_number_clusters+1)
plot.plot(range0, hierarchical_distortion_result, '-r', label=label_hierar)
plot.plot(range0, kmeans_distortion_result, '-g', label=label_kmeans)
plot.title('Hierarchical vs K-means distortion - PyCharm (IntelliJ)')
plot.xlabel('Size of Cluster')
plot.ylabel('Distortion')
plot.legend(loc='upper right')
plot.tight_layout()
if file_to_save:
file = file_to_save[:-4] + "_" + counties + '.png'
plot.savefig(file)
plot.clf()
def question6(URL, file_to_save, number_clusters=9, iterations=5, centers=False):
'''
:return:
'''
data_table, singleton_list = alg_project3_viz.run_example(URL)
cluster_list = kmeans_clustering(singleton_list, number_clusters, iterations)
print "Displaying", len(cluster_list), "k-means clusters"
if centers:
file = file_to_save[:-4] + 'with_centers' + '.png'
alg_clusters_matplotlib.plot_clusters(data_table, cluster_list, file, True)
alg_clusters_matplotlib.plot_clusters(data_table, cluster_list, file_to_save, False)
def question2(URL, file_to_save, number_clusters=15, centers=False):
'''
:return:
'''
data_table, singleton_list = alg_project3_viz.run_example(URL)
cluster_list = hierarchical_clustering(singleton_list, number_clusters)
print "Displaying", len(cluster_list), "hierarchical clusters"
if centers:
file = file_to_save[:-4] + 'with_centers' + '.png'
alg_clusters_matplotlib.plot_clusters(data_table, cluster_list, file, centers)
alg_clusters_matplotlib.plot_clusters(data_table, cluster_list, file_to_save, centers)
def question7(URL, number_clusters=9, iterations=5):
'''
:return:
'''
data_table, singleton_list = alg_project3_viz.run_example(URL)
cluster_list_hierarchical = hierarchical_clustering(singleton_list, number_clusters)
cluster_list_kmeans = kmeans_clustering(singleton_list, number_clusters, iterations)
hierarchical_distortion = sum([x.cluster_error(data_table) for x in cluster_list_hierarchical])
kmeans_distortion = sum([y.cluster_error(data_table) for y in cluster_list_kmeans])
counties = re.search('_(\d+)\.csv', URL).group(1)
print "Displaying distortion of hierarchical clusteringfor", counties, "counties", hierarchical_distortion
print "Displaying distortion of k-means clustering", counties, "counties", kmeans_distortion
def gen_random_clusters():
listx = []
listy_slow = []
listy_fast = []
data_table = alp.load_data_table(alp.DATA_896_URL)
for i in range(6,21):
list1 = []
sumerror = 0
sumerror_fast = 0
cluster_list = alp.run_example(i, True)
cluster_list_fast = alp.run_example(i, False)
for s in cluster_list:
error = s.cluster_error(data_table)
list1.append(error)
sumerror = sum(list1)
list1 = []
for s in cluster_list_fast:
error = s.cluster_error(data_table)
list1.append(error)
sumerror_fast = sum(list1)
listx.append(i)
listy_slow.append(sumerror)
listy_fast.append(sumerror_fast)
plt.plot(listx, listy_slow, '-r', label='hierarchical clustering')
plt.plot(listx, listy_fast, '-b', label='kmeans clustering')
plt.legend(loc='upper right')
plt.title("Quality - Data Set of 896")
plt.ylabel('Total error')
plt.xlabel('Number of clusters')
plt.show()