subgroup = X[allocation == 0]
plt.plot(subgroup[:, 0],
subgroup[:, 1],
color='r',
marker='x',
linestyle='None')
subgroup2 = X[allocation == 1]
plt.plot(subgroup2[:, 0],
subgroup2[:, 1],
color='b',
marker='.',
linestyle='None',
fillstyle='none')
plt.title('Spectral Clustering \n nmi against label : ' + str(spectral_nmi))
result = drc(X, 2, 0.5)
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, labels)
print 'DRC :\n\t', result['allocation']
print 'Dimension :\n', result['L']
plt.subplot(224)
subgroup = X[allocation == 0]
plt.plot(subgroup[:, 0],
subgroup[:, 1],
color='r',
marker='x',
linestyle='None')
subgroup2 = X[allocation == 1]
plt.text(3,
1,
str(result['L']),
spectral_nmi = normalized_mutual_info_score(allocation, univ_label)
print 'Spectral Clustering : ' , spectral_nmi
gmm = GMM(n_components=k)
gmm.fit(X)
allocation = gmm.predict(X)
gmm_nmi = normalized_mutual_info_score(allocation, univ_label)
print 'GMM : ' , gmm_nmi
##import pdb; pdb.set_trace()
#p = [0.3,0.4,0.5,0.6,0.7] #,0.8,0.9,1,1.1,1.2,1.3,1.4
#drc_nmi = 0
#for m in p:
result = drc(X, k, 1*Gamma, 1.1) # 1.1 turned out to be the best ratio
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, univ_label)
topic_nmi_new = normalized_mutual_info_score(allocation, topic_label)
print 'DRC : '
print '\tUniv : ' , drc_nmi
print '\tTopic: ' , topic_nmi_new
#print 'DRC : ' , m, ' , ' , drc_nmi_new
#if(drc_nmi < drc_nmi_new):
# drc_nmi = drc_nmi_new
#print 'Dimension :\n' , result['L']
objects = ('K means', 'Spectral', 'GMM', 'DRC')
y_pos = np.arange(len(objects))
clf = SpectralClustering(n_clusters=k, gamma=0.5)
allocation = clf.fit_predict(X)
spectral_nmi = normalized_mutual_info_score(allocation, labels)
print 'Pure Spectral Clustering :\n\t' ,allocation
plt.subplot(223)
subgroup = X[allocation == 0]
plt.plot(subgroup[:,0], subgroup[:,1], color='r' , marker='x', linestyle='None')
subgroup2 = X[allocation == 1]
plt.plot(subgroup2[:,0], subgroup2[:,1], color='b' , marker='.', linestyle='None', fillstyle='none')
plt.title('Spectral Clustering \n nmi against label : ' + str(spectral_nmi))
result = drc(X, 2, 0.5)
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, labels)
print 'DRC :\n\t' , result['allocation']
print 'Dimension :\n' , result['L']
plt.subplot(224)
subgroup = X[allocation == 0]
plt.plot(subgroup[:,0], subgroup[:,1], color='r' , marker='x', linestyle='None')
subgroup2 = X[allocation == 1]
plt.text(3, 1, str(result['L']), style='italic', bbox={'facecolor':'red', 'alpha':0.5, 'pad':10})
plt.plot(subgroup2[:,0], subgroup2[:,1], color='b' , marker='.', linestyle='None', fillstyle='none' )
plt.title('DRC / nmi against label : ' + str(drc_nmi))
plt.show()
clf = SpectralClustering(n_clusters=k, gamma=0.003)
allocation = clf.fit_predict(X)
spectral_nmi = normalized_mutual_info_score(allocation, label)
print 'Spectral Clustering : ', spectral_nmi
gmm = GMM(n_components=k)
gmm.fit(X)
allocation = gmm.predict(X)
gmm_nmi = normalized_mutual_info_score(allocation, label)
print 'GMM : ', gmm_nmi
#import pdb; pdb.set_trace()
#p = [0.001, 0.002, 0.003, 0.004, 0.005, 0.006, 0.007,0.008,0.009,0.010,0.011,0.012,0.013, 0.1, 0.2]
result = drc(X, 2, 0.8)
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, label)
print 'DRC : ', drc_nmi
#print 'Dimension :\n' , result['L']
#
#
#objects = ('K means', 'Spectral', 'DRC')
#y_pos = np.arange(len(objects))
#performance = [kmeans_nmi, spectral_nmi, drc_nmi]
#
#plt.bar(y_pos, performance, align='center', alpha=0.5)
#matplotlib.rc('xtick', labelsize=10)
#plt.xticks(y_pos, objects, rotation='vertical')
#plt.ylabel('NMI')
plt.subplot(223)
subgroup = X[allocation == 0]
plt.plot(subgroup[:, 0],
subgroup[:, 1],
color='r',
marker='x',
linestyle='None')
subgroup2 = X[allocation == 1]
plt.plot(subgroup2[:, 0],
subgroup2[:, 1],
color='b',
marker='.',
linestyle='None')
plt.title('Spectral Clustering \n nmi against label : ' + str(spectral_nmi))
result = drc(X, 2, 0.06)
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, labels)
#print 'DRC :\n\t' , result['allocation']
#print 'Dimension :\n' , result['L']
plt.subplot(224)
subgroup = X[allocation == 0]
plt.plot(subgroup[:, 0],
subgroup[:, 1],
color='r',
marker='x',
linestyle='None')
subgroup2 = X[allocation == 1]
plt.text(2,
0,
plt.plot(subgroup2[:,0], subgroup2[:,1], color='b' , marker='.', linestyle='None')
plt.title('Kmeans \n nmi against label : ' + str(kmeans_nmi))
clf = SpectralClustering(n_clusters=k, gamma=0.08)
allocation = clf.fit_predict(X)
spectral_nmi = normalized_mutual_info_score(allocation, labels)
plt.subplot(223)
subgroup = X[allocation == 0]
plt.plot(subgroup[:,0], subgroup[:,1], color='r' , marker='x', linestyle='None')
subgroup2 = X[allocation == 1]
plt.plot(subgroup2[:,0], subgroup2[:,1], color='b' , marker='.', linestyle='None')
plt.title('Spectral Clustering \n nmi against label : ' + str(spectral_nmi))
result = drc(X, 2, 0.06)
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, labels)
#print 'DRC :\n\t' , result['allocation']
#print 'Dimension :\n' , result['L']
plt.subplot(224)
subgroup = X[allocation == 0]
plt.plot(subgroup[:,0], subgroup[:,1], color='r' , marker='x', linestyle='None')
subgroup2 = X[allocation == 1]
plt.text(2, 0, str(np.round(result['L'],2)), style='italic', bbox={'facecolor':'red', 'alpha':0.5, 'pad':10})
plt.plot(subgroup2[:,0], subgroup2[:,1], color='b' , marker='.', linestyle='None')
plt.title('DRC \n nmi against label : ' + str(drc_nmi))
plt.show()
spectral_nmi = normalized_mutual_info_score(allocation, label)
print 'Spectral Clustering : ' , spectral_nmi
gmm = GMM(n_components=k)
gmm.fit(X)
allocation = gmm.predict(X)
gmm_nmi = normalized_mutual_info_score(allocation, label)
print 'GMM : ' , gmm_nmi
#import pdb; pdb.set_trace()
#p = [0.001, 0.002, 0.003, 0.004, 0.005, 0.006, 0.007,0.008,0.009,0.010,0.011,0.012,0.013, 0.1, 0.2]
result = drc(X, 2, 0.8)
allocation = result['allocation']
drc_nmi = normalized_mutual_info_score(allocation, label)
print 'DRC : ' , drc_nmi
#print 'Dimension :\n' , result['L']
#
#
#objects = ('K means', 'Spectral', 'DRC')
#y_pos = np.arange(len(objects))
#performance = [kmeans_nmi, spectral_nmi, drc_nmi]
#
#plt.bar(y_pos, performance, align='center', alpha=0.5)
#matplotlib.rc('xtick', labelsize=10)
#plt.xticks(y_pos, objects, rotation='vertical')
d_matrix = sklearn.metrics.pairwise.pairwise_distances(X,
Y=None,
metric='euclidean')
sigma = np.median(d_matrix)
Gamma = 1 / (2 * np.power(sigma, 2))
#clf = SpectralClustering(n_clusters=k, gamma=Gamma)
#allocation = clf.fit_predict(X)
#spectral_nmi = normalized_mutual_info_score(allocation, univ_label)
#print 'Spectral Clustering : ' , spectral_nmi
#gmm = GMM(n_components=k)
#gmm.fit(X)
#allocation = gmm.predict(X)
result = drc(X, k, 0.07 * Gamma,
Const=50.1) # 1.1 turned out to be the best ratio
allocation = result['allocation']
if True: # Plot clustering results
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
Uq_a = np.unique(allocation)
group1 = X[allocation == Uq_a[0]]
group2 = X[allocation == Uq_a[1]]
group3 = X[allocation == Uq_a[2]]
ax.scatter(group1[:, 0], group1[:, 1], group1[:, 2], c='b', marker='o')
ax.scatter(group2[:, 0], group2[:, 1], group2[:, 2], c='r', marker='x')
ax.scatter(group3[:, 0], group3[:, 1], group3[:, 2], c='g', marker='x')
ax.set_xlabel('Feature 1')
sigma = np.median(d_matrix) Gamma = 1/(2*np.power(sigma,2)) #clf = SpectralClustering(n_clusters=k, gamma=Gamma) #allocation = clf.fit_predict(X) #spectral_nmi = normalized_mutual_info_score(allocation, univ_label) #print 'Spectral Clustering : ' , spectral_nmi #gmm = GMM(n_components=k) #gmm.fit(X) #allocation = gmm.predict(X) result = drc(X, k, 0.07*Gamma, Const=50.1) # 1.1 turned out to be the best ratio allocation = result['allocation'] if True: # Plot clustering results fig = plt.figure() ax = fig.add_subplot(111, projection='3d') Uq_a = np.unique(allocation) group1 = X[allocation == Uq_a[0]] group2 = X[allocation == Uq_a[1]] group3 = X[allocation == Uq_a[2]] ax.scatter(group1[:,0], group1[:,1], group1[:,2], c='b', marker='o') ax.scatter(group2[:,0], group2[:,1], group2[:,2], c='r', marker='x') ax.scatter(group3[:,0], group3[:,1], group3[:,2], c='g', marker='x')