def _CCA(data, graph, n):
cca = CCA(n_components=n)
adjacencyMatrix = createAffinityMatrix(graph)
cca.fit(data, adjacencyMatrix)
X_c, Y_c = cca.transform(data, adjacencyMatrix)
writeCSV(X_c, 'CCA_X')
writeCSV(Y_c, 'CCA_Y')
def runClustering(train):
ac = AgglomerativeClustering(n_clusters=10).fit(train)
labels = ac.labels_
return labels
if __name__ == '__main__':
path = sys.argv[1]
output = sys.argv[2]
print 'Loading data...'
seed, similarity, train = loadData(path)
# print 'Creating connectivity matrix...'
# connectivity = createConnectivityMatrix(similarity)
print 'Running clustering algorithm...'
labels = runClustering(train)
print 'Creating CSV File...'
name = 'AgglomerativeLabels'
writeCSV(labels, name)
print 'Visualing clusters...'
result = np.array(
pd.read_csv("./Scripts/Results/" + name + '.csv', header=None))
cluster = assignClusters(seed, result)
pred = createPred(seed, cluster, result)
writeCSV(pred, output, ['Id', 'Label'])
def runSpectralClustering(train):
sc = SpectralClustering(n_clusters=10,
affinity='nearest_neighbors').fit(train)
labels = sc.labels_
return labels
if __name__ == '__main__':
start = time.time()
path = sys.argv[1]
print 'Loading data...'
seed, train = loadData(path)
print 'Running Spectral Clustering...'
labels = runSpectralClustering(train)
print 'Creating CSV File...'
name = 'SCLabels'
writeCSV(labels, name)
print 'Visualing clusters...'
result = np.array(
pd.read_csv("./Scripts/Results/" + name + '.csv', header=None))
cluster = assignClusters(seed, result)
# pred = createPred(seed, cluster, result)
# writeCSV(pred[6000:], 'SCPred', ['Id', 'Label'])
print time.time() - start
def combineLabels2(data, seed):
label0 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(0).csv", header=None))
assignment0 = makeAssignment(label0)
print '0 - ' + str(len(assignment0))
label1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(1).csv", header=None))
assignment1 = makeAssignment(label1)
print '1 - ' + str(len(assignment1))
label2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(2).csv", header=None))
assignment2 = makeAssignment(label2)
print '2 - ' + str(len(assignment2))
label3_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(3)1.csv", header=None))
label3_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(3)2.csv", header=None))
assignment3 = {}
for x, y in label3_1:
l = assignment3.setdefault(x, [])
l.append(y)
assignment3[x] = l
for x, y in label3_2:
l = assignment3.setdefault(x, [])
l.append(y)
assignment3[x] = l
for key, val in assignment3.items():
if len(val) < 2:
del assignment3[key]
print '3 - ' + str(len(assignment3))
label4_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(4)1.csv", header=None))
label4_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(4)2.csv", header=None))
assignment4 = {}
for x, y in label4_1:
l = assignment4.setdefault(x, [])
l.append(y)
assignment4[x] = l
for x, y in label4_2:
l = assignment4.setdefault(x, [])
l.append(y)
assignment4[x] = l
for key, val in assignment4.items():
if len(val) < 2:
del assignment4[key]
print '4 - ' + str(len(assignment4))
label5_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(5)1.csv", header=None))
label5_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(5)2.csv", header=None))
assignment5 = {}
for x, y in label5_1:
l = assignment5.setdefault(x, [])
l.append(y)
assignment5[x] = l
for x, y in label5_2:
l = assignment5.setdefault(x, [])
l.append(y)
assignment5[x] = l
for key, val in assignment5.items():
if len(val) < 2:
del assignment5[key]
print '5 - ' + str(len(assignment5))
label6 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(6).csv", header=None))
assignment6 = makeAssignment(label6)
print '6 - ' + str(len(assignment6))
label7_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(7)1.csv", header=None))
label7_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(7)2.csv", header=None))
assignment7 = {}
for x, y in label7_1:
l = assignment7.setdefault(x, [])
l.append(y)
assignment7[x] = l
for x, y in label7_2:
l = assignment7.setdefault(x, [])
l.append(y)
assignment7[x] = l
for key, val in assignment7.items():
if len(val) < 2:
del assignment7[key]
print '7 - ' + str(len(assignment7))
label8_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(8)1.csv", header=None))
label8_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(8)2.csv", header=None))
assignment8 = {}
for x, y in label8_1:
l = assignment8.setdefault(x, [])
l.append(y)
assignment8[x] = l
for x, y in label8_2:
l = assignment8.setdefault(x, [])
l.append(y)
assignment8[x] = l
for key, val in assignment8.items():
if len(val) < 2:
del assignment8[key]
print '8 - ' + str(len(assignment8))
label9_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(9)1.csv", header=None))
label9_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(9)2.csv", header=None))
assignment9 = {}
for x, y in label9_1:
l = assignment9.setdefault(x, [])
l.append(y)
assignment9[x] = l
for x, y in label9_2:
l = assignment9.setdefault(x, [])
l.append(y)
assignment9[x] = l
for key, val in assignment9.items():
if len(val) < 2:
del assignment9[key]
print '9 - ' + str(len(assignment9))
assignments = [
assignment0, assignment1, assignment2, assignment3, assignment4,
assignment5, assignment6, assignment7, assignment8, assignment9
]
assignment = {}
for i in range(len(data)):
for digit in range(10):
if i + 1 in assignments[digit]:
assignment[i + 1] = digit
for x, y in seed:
assignment[x] = y
d = {}
res = []
for key, val in assignment.items():
res.append([key, val])
count = d.setdefault(val, 0)
d[val] = count + 1
print d
writeCSV(res, 'assignment5')
def combineLabels(data):
label0 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(0).csv", header=None))
assignment0 = makeAssignment(label0)
print '0 - ' + str(len(assignment0))
label1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(1).csv", header=None))
assignment1 = makeAssignment(label1)
print '1 - ' + str(len(assignment1))
label2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(2).csv", header=None))
assignment2 = makeAssignment(label2)
print '2 - ' + str(len(assignment2))
label3 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(3).csv", header=None))
assignment3 = makeAssignment(label3)
print '3 - ' + str(len(assignment3))
label6 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(6).csv", header=None))
assignment6 = makeAssignment(label6)
print '6 - ' + str(len(assignment6))
label5_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(5)1.csv", header=None))
label5_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(5)2.csv", header=None))
label5_3 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(5)3.csv", header=None))
print 'Assignment 5'
assignment5 = {}
for x, y in label5_1:
l = assignment5.setdefault(x, [])
l.append(y)
assignment5[x] = l
for x, y in label5_2:
l = assignment5.setdefault(x, [])
l.append(y)
assignment5[x] = l
for x, y in label5_3:
l = assignment5.setdefault(x, [])
l.append(y)
assignment5[x] = l
print '5 - ' + str(len(assignment5))
# for key, val in assignment5.items():
# if len(val) < 2:
# del assignment5[key]
# print len(assignment5)
label7_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(7).csv", header=None))
label7_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(7)1.csv", header=None))
label7_3 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(7)2.csv", header=None))
label7_4 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(7)3.csv", header=None))
print 'Assignment 7'
assignment7 = {}
for x, y in label7_1:
l = assignment7.setdefault(x, [])
l.append(y)
assignment7[x] = l
for x, y in label7_2:
l = assignment7.setdefault(x, [])
l.append(y)
assignment7[x] = l
for x, y in label7_3:
l = assignment7.setdefault(x, [])
l.append(y)
assignment7[x] = l
for x, y in label7_4:
l = assignment7.setdefault(x, [])
l.append(y)
assignment7[x] = l
print len(assignment7)
for key, val in assignment7.items():
if len(val) < 3:
del assignment7[key]
print '7 - ' + str(len(assignment7))
label8_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(8)1.csv", header=None))
label8_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(8)2.csv", header=None))
label8_3 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(8)3.csv", header=None))
print 'Assignment 8'
assignment8 = {}
for x, y in label8_1:
l = assignment8.setdefault(x, [])
l.append(y)
assignment8[x] = l
for x, y in label8_2:
l = assignment8.setdefault(x, [])
l.append(y)
assignment8[x] = l
for x, y in label8_3:
l = assignment8.setdefault(x, [])
l.append(y)
assignment8[x] = l
for key, val in assignment8.items():
if len(val) < 2:
del assignment8[key]
print '8 - ' + str(len(assignment8))
label4_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(4)1.csv", header=None))
label4_2 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(4)2.csv", header=None))
label9_1 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(9)1.csv", header=None))
label49 = np.array(
pd.read_csv("./Scripts/Results/TrainLabel_(49).csv", header=None))
print 'Assignment 4 and 9'
assignment49 = {}
for x, y in label4_1:
l = assignment49.setdefault(x, [])
l.append(y)
assignment49[x] = l
for x, y in label4_2:
l = assignment49.setdefault(x, [])
l.append(y)
assignment49[x] = l
for x, y in label9_1:
l = assignment49.setdefault(x, [])
l.append(y)
assignment49[x] = l
for x, y in label49:
l = assignment49.setdefault(x, [])
l.append(y)
assignment49[x] = l
assignment4 = {}
assignment9 = {}
for key, val in assignment49.items():
if val.count(4) == 2 and val.count(9) == 2:
a = random.randint(0, 1)
if a == 0:
val.append(4)
else:
val.append(9)
if val.count(4) >= 2:
l = assignment4.setdefault(key, [])
l.append(val)
assignment4[key] = l
elif val.count(9) >= 2:
l = assignment9.setdefault(key, [])
l.append(val)
assignment9[key] = l
print '4 - ' + str(len(assignment4))
print '9 - ' + str(len(assignment9))
assignments = [
assignment0, assignment1, assignment2, assignment3, assignment4,
assignment5, assignment6, assignment7, assignment8, assignment9
]
assignment = {}
for i in range(len(data)):
for digit in range(10):
if i + 1 in assignments[digit]:
assignment[i + 1] = digit
d = {}
res = []
for key, val in assignment.items():
res.append([key, val])
count = d.setdefault(val, 0)
d[val] = count + 1
writeCSV(res, 'assignment3')