def main(max, classifyMetric):
for k in range(1, max + 1):
foutModel = open(
path + str(k) + "_" + X + "_" + Y + "_" + classifyMetric +
"_result_2.txt", "w")
knn.main(sys.argv[1], k, foutModel, classifyMetric)
foutModel.close
def main():
num_hiddens = 10
eps = 0.1
momentum = 0.0
num_epochs = 1000
current_problem = [2.1, 2.2, 2.3, 2.4, 2.5]
print "Running problems: ", current_problem
# 2.1 and 2.2
if 2.2 in current_problem and 2.1 in current_problem:
W1, W2, b1, b2, train_error, valid_error, train_class_error, valid_class_error = TrainNN(num_hiddens, eps, momentum, num_epochs)
DisplayErrorPlot(train_error, valid_error, mode='cross_entropy')
DisplayErrorPlot(train_class_error, valid_class_error, mode='classification_error')
# 2.3
if 2.3 in current_problem:
for eps in [0.5, 0.2, 0.01]:
W1, W2, b1, b2, train_error, valid_error, train_class_error, valid_class_error = TrainNN(num_hiddens, eps, momentum, num_epochs)
# iterate through different eps
suffix = '_at_eps_' + str(eps)
suffix = suffix.replace('.', ',')
DisplayErrorPlot(train_error, valid_error, mode='cross_entropy' + suffix)
DisplayErrorPlot(train_class_error, valid_class_error, mode='classification_error' + suffix)
eps = 0.1
for momentum in [0.9, 0.5, 0.0]:
W1, W2, b1, b2, train_error, valid_error, train_class_error, valid_class_error = TrainNN(num_hiddens, eps, momentum, num_epochs)
# iterate through different momentum
suffix = '_at_momentum_' + str(momentum)
suffix = suffix.replace('.', ',')
DisplayErrorPlot(train_error, valid_error, mode='cross_entropy' + suffix)
DisplayErrorPlot(train_class_error, valid_class_error, mode='classification_error' + suffix)
# 2.4
if 2.4 in current_problem:
eps = 0.02
momentum = 0.5
for num_hiddens in [2, 5, 10, 30, 100]:
start = time.time()
(
W1, W2, b1, b2,
train_error, valid_error, test_error,
train_class_error, valid_class_error, test_class_error,
) = TrainNN(num_hiddens, eps, momentum, num_epochs, run_test=True)
elapsed = time.time() - start
print "Training time at hidden_unit=", num_hiddens, " is ", elapsed, "seconds"
# iterate through different num_hiddens
suffix = '_at_hidden_unit_' + str(num_hiddens)
DisplayErrorPlot(train_error, valid_error, mode='cross_entropy' + suffix, test=test_error)
DisplayErrorPlot(train_class_error, valid_class_error, mode='classification_error' + suffix, test=test_class_error)
if 2.5 in current_problem:
import knn
knn.main()
def main(testfile=None): if testfile: a=rd.read(testfile) k=3 predictions=[] for x in range(len(a)): neighbors = knn.getNeighbors(train, a[x], k) result = knn.getResponse(neighbors) predictions.append(result) return(predictions) else: knn.main(train,tests)
def main(i=0, j=62, testfile=None):
tr, ts = rt.read2(i, j)
k = 3
if testfile:
tst = rd.read2(testfile)
predictions = []
for x in range(len(tst)):
neighbors = knn.getNeighbors(tr, tst[x], k)
result = knn.getResponse(neighbors)
predictions.append(result)
return (predictions)
else:
knn.main(tr, ts)
def main(kInKnn):
start = time.time()
[trainingX, trainingY, testX, testY] = loadDataset("ex2data1train.csv", "ex2data1test.csv")
pred = knn.main(trainingX, trainingY, testX, kInKnn)
end = time.time()
count = 0
for i in range(len(pred)):
if pred[i] == testY[i]:
count += 1
print("accuracy: " + str(round(count / float(len(testY)) * 100, 2)) + "%")
print("training time: " + str(round(end - start, 1)) + " seconds")
def main(kInKnn):
start = time.time()
[trainingX, trainingY, testX, testY] = loadDataset('ex2data1train.csv',
'ex2data1test.csv')
pred = knn.main(trainingX, trainingY, testX, kInKnn)
end = time.time()
count = 0
for i in range(len(pred)):
if pred[i] == testY[i]:
count += 1
print('accuracy: ' + str(round(count / float(len(testY)) * 100, 2)) + '%')
print('training time: ' + str(round(end - start, 1)) + ' seconds')
def bagging(N):
#training data
xRF = RFData.x
y = RFData.y
#testing data
xtRF = RFData.xt
yt = RFData.yt
xKNN = knnData.main()[0] #training X
xtKNN = knnData.main()[2] #test X
countYPredict = []
for i in range(len(yt)):
countYPredict.append(0)
for k in range(N): # number of bootstrapping
x_RF = []
y_RF = []
x_KNN = []
y_KNN = []
# bootstrapping
for i in range(int(len(xRF) * 0.6)):
r = randint(0, len(xRF) - 1)
tRF = []
for j in range(len(xRF[0]) - 1):
tRF.append(xRF[r][j])
# for RF, data duplicates are not allowed
if tRF not in x_RF:
x_RF.append(tRF)
y_RF.append(y[r])
x_KNN.append(xKNN[r])
y_KNN.append(y[r])
# RF
start = time.time()
rf = RF(B=TreeNum, Bagging=isBagging)
rf.train(x_RF, y_RF)
pred = rf.predict(xtRF)
end = time.time()
count = 0
for i in range(len(pred)):
if pred[i] == yt[i]:
count += 1
print("RF, trial #" + str(k + 1) + ": ")
print('\taccuracy: ' + str(round(count / float(len(yt)) * 100, 2)) +
'%')
print('\ttraining time: ' + str(round(end - start, 1)) + ' seconds')
for i in range(len(pred)):
countYPredict[i] = countYPredict[i] + pred[i]
# KNN
start = time.time()
pred = knn.main(x_KNN, y_KNN, xtKNN, kInKnn)
end = time.time()
count = 0
for i in range(len(pred)):
if pred[i] == yt[i]:
count += 1
print("KNN, trial #" + str(k + 1) + ": ")
print('\taccuracy: ' + str(round(count / float(len(yt)) * 100, 2)) +
'%')
print('\ttraining time: ' + str(round(end - start, 1)) + ' seconds')
for i in range(len(pred)):
countYPredict[i] = countYPredict[i] + pred[i]
finalPredict = []
for i in range(len(yt)):
if countYPredict[i] >= N:
finalPredict.append(1)
else:
finalPredict.append(0)
count = 0
for i in range(len(finalPredict)):
if finalPredict[i] == yt[i]:
count += 1
print()
print('After combining the classifiers by bagging: ')
print('\taccuracy: ' + str(round(count / float(len(yt)) * 100, 2)) + '%')
while choice2 != 4:
print('\nkNN Classification:')
print(
'1 -> Classify Dataset. (Splits Dataset into training and test set)'
)
print(
'2 -> Classify Testset. (Test kNN accuracy using Dataset as training set)'
)
print('3 -> Classify New Emails.\n4 -> Back.')
choice2 = int(input('Enter your choice: '))
if choice2 == 1:
# Classify Dataset
dataset_name = input(
'Enter the name of dataset folder to be classified: ')
knn.main(dataset_name)
elif choice2 == 2:
# Classify Testset
dataset_name = input(
'Enter the name of dataset folder to be used as training set: '
)
testset_name = input('Enter the name of test set folder: ')
knn_classify.main(dataset_name, testset_name)
elif choice2 == 3:
# Classify New Emails
dataset_name = input(
'Enter the name of dataset folder to be used as training set: '
)
reply = input(
def hello():
in_data = request.json['details']
predictions = main(in_data)
return predictions
def main(argv):
arg_index_start = 0
DEBUG = "FALSE"
try:
opts, args = getopt.getopt(argv, 'd', ['debug'])
if not args:
usage()
sys.exit(2)
except getopt.GetoptError as err:
usage()
sys.exit(2)
for opt, arg in opts:
if opt in ('-d'):
DEBUG = "TRUE"
arg_index_start = 1
# arguments
if not argv[arg_index_start + 4]:
usage()
sys.exit(2)
if os.path.isdir(argv[arg_index_start]):
feature_files_dir = argv[arg_index_start]
else:
print("Feature directory does not exist: ",
argv[arg_index_start],
file=sys.stderr)
usage()
sys.exit(2)
if os.path.isfile(argv[arg_index_start + 1]):
outtype_dataset_file = argv[arg_index_start + 1]
else:
print("Out type datafile does not exist: ",
argv[arg_index_start + 1],
file=sys.stderr)
usage()
sys.exit(2)
i = 2
datatype_filenames = []
datatypes = []
dist_metrics = []
weights = []
while i < len(args):
if os.path.isfile(argv[arg_index_start + i]):
datatype_filename = argv[arg_index_start + i]
datatype_filenames.append(datatype_filename)
datatype = os.path.basename(datatype_filename.rsplit('-', 1)[0])
datatypes.append(datatype)
else:
print("Input datatype file does not exist: ",
argv[arg_index_start + i],
file=sys.stderr)
usage()
sys.exit(2)
if argv[arg_index_start + (i + 1)]:
dist_metrics.append(argv[arg_index_start + (i + 1)])
else:
print("No distance metric provided for ",
argv[arg_index_start + i],
file=sys.stderr)
usage()
sys.exit(2)
if argv[arg_index_start + (i + 2)]:
weights.append(argv[arg_index_start + (i + 2)])
else:
print("No weight provided for ",
argv[arg_index_start + i],
file=sys.stderr)
usage()
sys.exit(2)
i = i + 3
outtype_col_name = "Id"
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": datatypes:",
datatypes,
file=sys.stderr)
print("*** DEBUG: " + sys.argv[0] + ": dist_metrics:",
dist_metrics,
file=sys.stderr)
print("*** DEBUG: " + sys.argv[0] + ": weights:",
weights,
file=sys.stderr)
total_ids_scores = []
for datatype_num in range(len(datatypes)):
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": datatype_num:",
datatype_num,
file=sys.stderr)
features_file = feature_files_dir + "/" + datatypes[
datatype_num] + ".tmp"
if not os.path.isfile(features_file):
print("Features file does not exist: ",
features_file,
file=sys.stderr)
sys.exit(2)
knn_args = features_file + " " + outtype_dataset_file + " " + datatype_filenames[
datatype_num] + " " + dist_metrics[datatype_num] + " false"
ids_scores = knn.main(knn_args.split())
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": ids_scores:",
ids_scores,
file=sys.stderr)
total_ids_scores = total_ids_scores + ids_scores
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": total_ids_scores:",
total_ids_scores,
file=sys.stderr)
new_total_ids_scores = []
for id_score in total_ids_scores:
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": id_score:",
id_score,
file=sys.stderr)
found = "FALSE"
for new_id_score in new_total_ids_scores:
if new_id_score[0] == id_score[0]:
new_id_score[1] = new_id_score[1] + id_score[1]
found = "TRUE"
break
if found == "FALSE":
new_total_ids_scores.append(id_score)
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": new_total_ids_scores:",
new_total_ids_scores,
file=sys.stderr)
for id_score in new_total_ids_scores:
id_score[1] = id_score[1] / len(datatypes)
new_total_ids_scores.sort(key=lambda x: x[1], reverse=True)
if DEBUG == "TRUE":
print("*** DEBUG: " + sys.argv[0] + ": new_total_ids_score:",
new_total_ids_scores,
file=sys.stderr)
return (new_total_ids_scores)
def bagging(N):
# training data
xRF = RFData.x
y = RFData.y
# testing data
xtRF = RFData.xt
yt = RFData.yt
xKNN = knnData.main()[0] # training X
xtKNN = knnData.main()[2] # test X
countYPredict = []
for i in range(len(yt)):
countYPredict.append(0)
for k in range(N): # number of bootstrapping
x_RF = []
y_RF = []
x_KNN = []
y_KNN = []
# bootstrapping
for i in range(int(len(xRF) * 0.6)):
r = randint(0, len(xRF) - 1)
tRF = []
for j in range(len(xRF[0]) - 1):
tRF.append(xRF[r][j])
# for RF, data duplicates are not allowed
if tRF not in x_RF:
x_RF.append(tRF)
y_RF.append(y[r])
x_KNN.append(xKNN[r])
y_KNN.append(y[r])
# RF
start = time.time()
rf = RF(B=TreeNum, Bagging=isBagging)
rf.train(x_RF, y_RF)
pred = rf.predict(xtRF)
end = time.time()
count = 0
for i in range(len(pred)):
if pred[i] == yt[i]:
count += 1
print("RF, trial #" + str(k + 1) + ": ")
print("\taccuracy: " + str(round(count / float(len(yt)) * 100, 2)) + "%")
print("\ttraining time: " + str(round(end - start, 1)) + " seconds")
for i in range(len(pred)):
countYPredict[i] = countYPredict[i] + pred[i]
# KNN
start = time.time()
pred = knn.main(x_KNN, y_KNN, xtKNN, kInKnn)
end = time.time()
count = 0
for i in range(len(pred)):
if pred[i] == yt[i]:
count += 1
print("KNN, trial #" + str(k + 1) + ": ")
print("\taccuracy: " + str(round(count / float(len(yt)) * 100, 2)) + "%")
print("\ttraining time: " + str(round(end - start, 1)) + " seconds")
for i in range(len(pred)):
countYPredict[i] = countYPredict[i] + pred[i]
finalPredict = []
for i in range(len(yt)):
if countYPredict[i] >= N:
finalPredict.append(1)
else:
finalPredict.append(0)
count = 0
for i in range(len(finalPredict)):
if finalPredict[i] == yt[i]:
count += 1
print()
print("After combining the classifiers by bagging: ")
print("\taccuracy: " + str(round(count / float(len(yt)) * 100, 2)) + "%")
import knn
import matplotlib.pyplot as plt
from pylab import *
k_values = [1,5,10,15,30,50]
p = 0.5
for k in k_values:
knn.main("ALL.dat", "AML.dat", k, p)
print "\n"
# plot(k_values, [0.93, 0.94, 0.83, 0.79, 0.61, 0.61])
# xlabel('k')
# ylabel('accuracy')
# savefig('question1.jpg')
print "\n\n---------------------------------\n\n"
p_values = [0, 0.05, 0.20, 0.50, 0.75, 0.95, 1.00]
results = []
k = 30
for p in p_values:
#knn.main("ALL.dat", "AML.dat", k, p)
print "\n"
fig, ax = plt.subplots()
sensitivity = [1.0, 1.0, 1.0, 1.0, 0.77, 0.0, 0.0]
specificity = [0.0, 0.0, 0.0, 0.0, 1.0, 1.0, 1.0]
to_plot_spec = [1 - x for x in specificity]
ax.plot(sensitivity, to_plot_spec)
print(
'\tk - [OPTIONAL] number of neighbours for knn subsection \'a\''
)
if __name__ == '__main__':
if len(sys.argv) <= 2:
usage()
sys.exit(1)
mod = sys.argv[1]
sub = sys.argv[2]
ret = 0
if mod == 'knn':
import knn
ret = knn.main(sub, sys.argv[3:])
elif mod == 'perceptron':
import perceptron
ret = perceptron.main(sub)
elif mod == 'svm':
import svm
ret = svm.main(sub)
else:
usage()
sys.exit(1)
if ret != 0:
usage()
print('Invalid Input.')
elif choice == 2:
# Classify using kNN
choice2 = 0
while choice2 != 4:
print('\nkNN Classification:')
print('1 -> Classify Dataset. (Splits Dataset into training and test set)')
print('2 -> Classify Testset. (Test kNN accuracy using Dataset as training set)')
print('3 -> Classify New Emails.\n4 -> Back.')
choice2 = int(input('Enter your choice: '))
if choice2 == 1:
# Classify Dataset
dataset_name = input('Enter the name of dataset folder to be classified: ')
knn.main(dataset_name)
elif choice2 == 2:
# Classify Testset
dataset_name = input('Enter the name of dataset folder to be used as training set: ')
testset_name = input('Enter the name of test set folder: ')
knn_classify.main(dataset_name, testset_name)
elif choice2 == 3:
# Classify New Emails
dataset_name = input('Enter the name of dataset folder to be used as training set: ')
reply = input('Do you want get new unread emails from your email account? (y/n): ')[0].lower()
if reply == 'y':
usr = input('Email: ')
pwd = getpass('Password: ')
def process_message(msg):
msg = msg.lower()
global qno
global ans
#NLTK functions
hlink = "nationalrail.co.uk"
label_link.configure(text=hlink, fg="blue", cursor="hand2")
label_link.bind("<Button-1>", openLink)
hiList = [
"hello", "hi", "hey", "heya", "hiya", "hai", "howdy", "ciao", "ni hao"
]
biList = [
"goodbye", "bye", "bi", "exit", "see you", "cya", "see ya", "byebye",
"ciao", "bye bye"
]
if msg.lower() in hiList:
return "Hello to you too user!"
elif msg.lower() in biList:
sys.exit()
if msg.lower() == "good":
return "That's very good"
delayList = ["delay", "late", "delayed"]
print(qno)
k = False
for w in delayList:
if msg in delayList:
k = True
if k == True:
if qno == 0 or qno > 4 and qno < 10:
qno = 10
qno += 1
print()
return "How long has your train been delayed by?"
if qno == 11:
msg = re.sub(r'[^\d]', "", msg)
msg = int(msg)
ans[0] = msg
qno += 1
return "Where did your train depart from and arriving at?"
elif qno == 12:
x = nlp.findStation(msg)
ke.startKE(x[1], x[0], "", "", "")
facts = readfacts()
if facts[0] != "null":
qno += 1
for s in stations:
if x[1] == s[0]:
x[1] = s[1]
if x[0] == s[0]:
x[0] = s[1]
ans[1] = x[1]
ans[2] = x[0]
return "What time did you depart?"
elif qno == 13:
x = nlp.findTime(msg)
x = re.sub(r'[^\d]', "", x)
ans[3] = int(x)
qno += 1
return "What was you orginal estimated arrival?"
elif qno == 14:
x = nlp.findTime(msg)
x = re.sub(r'[^\d]', "", x)
ans[4] = int(x)
qno += 1
return "OK 1 sec!"
elif qno == 15:
s1 = " ".join(ans[1])
s2 = " ".join(ans[2])
x = knn.main(ans[4], ans[3], stationsDict[s1], stationsDict[s2],
ans[0])
x = "The estimated time of arrival is " + x
resetfacts()
qno = 0
return x
if qno == 0:
x = nlp.findStation(msg)
facts = readfacts()
print(facts)
print("asdaf")
ke.startKE(x[1], x[0], "", "", "")
facts = readfacts()
print(facts)
if facts[0] != "null":
qno += 1
ans[0] = x[1]
if facts[1] != "null":
qno += 1
ans[1] = x[0]
return facts[5]
elif qno == 1:
x = nlp.finddplace(msg)
facts = readfacts()
ke.startKE(x, ans[0], "", "", "")
facts = readfacts()
if facts[0] != "null":
qno += 1
ans[0] = x
return facts[5]
elif qno == 2:
x = nlp.findDate(msg)
ke.startKE(ans[0], ans[1], "", x, "")
facts = readfacts()
if facts[2] != "null":
qno += 1
ans[2] = x
return facts[5]
elif qno == 3:
x = nlp.findTime(msg)
ke.startKE(ans[0], ans[1], x, ans[2], "")
facts = readfacts()
if facts[3] != "null":
qno += 1
ans[3] = x
return facts[5]
elif qno == 4:
x = nlp.checkDepart(msg)
ke.startKE(ans[0], ans[1], ans[3], ans[2], x)
facts = readfacts()
if facts[4] != "null":
qno += 1
ans[4] = x
temp = webScraper.main()
hlink = temp[1]
label_link.configure(text=hlink, fg="blue", cursor="hand2")
return temp[0]
return ["arrival or departure?"]
if msg == " ":
return "Hello?"
facts = readfacts()
print(facts)
# while "null" in facts:
#ke.test(msg)
# print("NUL IN FACT")
return "Sorry! I don't understand!"
frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
fgmask = fgbg.apply(frame)
fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_OPEN, kernel)
#fgmask = cv2.filter2D(fgmask,-1,kernel)
#fgmask = cv2.dilate(fgmask,kernel,iterations = 1)
fgmask = cv2.morphologyEx(fgmask, cv2.MORPH_CLOSE, kernel2)
res = cv2.bitwise_and(fgmask, mask)
cv2.imshow("frame", res)
k = cv2.waitKey(1) & 0xFF
#if k == ord('a'):
feature = cleaner.do(res)
if feature != None:
knn.main(float(feature[0]), float(feature[1]))
if k == ord('q'):
break
time.sleep(0.2)
cap.release()
cv2.destroyAllWindows()
with open('recent.file', 'w') as myfile:
myfile.write(feature_data)
print(feature_data)
while True:
# mengcapture frame realtime
(ret, frame) = cap.read()
cv2.putText(
frame,
'Prediksi: ' + prediction,
(15, 60),
cv2.FONT_HERSHEY_SIMPLEX,1, (0, 0, 255), 2)
# window output
cv2.imshow('Sistem Pendeteksi warna kematangan tomat', frame)
color_histogram_of_test_image(frame)
prediction = knn.main('training.csv', 'recent.file')
if cv2.waitKey(1) & 0xFF == ord('q'):
break
cap.release()
cv2.destroyAllWindows()
