predictions = []
idx = np.random.choice(n_test, n_samples, replace=False)
train_img_sample = np.array(face_data.face_train_imgs)[idx].tolist()
train_labels_sample = np.array(face_data.face_train_labels)[idx].tolist()
nbd = NaiveBayesPredictor(train_img_sample, train_labels_sample, range(2))
for feature in test_features:
predictions.append(nbd.predict(feature))
correct, wrong = (0, 0)
for k, pred in enumerate(predictions):
if pred == face_data.face_test_labels[k]:
correct += 1
else:
wrong += 1
print("Stats while using {} % of training data".format(n * 10))
# print("The predictions are: ", predictions)
# print("The actual labels are:", face_data.face_test_labels)
print("No. of correct guesses = {}".format(correct))
print("No. of wrong guesses = {}".format(wrong))
percentage = (correct * 100) / (correct + wrong)
print("Percentage accuracy: {}".format(percentage))
percentage_list.append(percentage)
run_time = time.time() - start_time
print("The run_time in seconds is: {}".format(run_time))
time_list.append(run_time)
p3_utils.plot_line_graph(range(10, 101, 10), percentage_list, "Naive Bayes Accuracy chart for faces",
"Percentage of training data used", "Accuracy obtained on test data")
p3_utils.plot_line_graph(range(10, 101, 10), time_list, "Naive Bayes Runtime chart for faces",
"Percentage of training data used", "Run time in seconds")
print(statistics.stdev(percentage_list))
exec_time = []
accuracy = []
for i in np.arange(0.1, 1.1, 0.1):
predictions = []
knnf = KnnFaces(5, i)
# Run this with a pool of 10 agents having a chunksize of 10 until finished
agents = 10
chunksize = 10
with Pool(processes=agents) as pool:
predictions = pool.map(knnf.predict, knnf.testData, chunksize)
# cool_visualization(digit_data)
correct, wrong = (0, 0)
for k, pred in enumerate(predictions):
if pred == knnf.face_data.face_test_labels[k]:
correct += 1
else:
wrong += 1
print("The predictions are: ", predictions)
print("The actual labels are:", knnf.face_data.face_test_labels)
print("No. of correct guesses = {}".format(correct))
print("No. of wrong guesses = {}".format(wrong))
print("Percentage accuracy: {}".format((correct * 100) / (correct + wrong)))
accuracy.append((correct * 100) / (correct + wrong))
print('execution time', time.time() - start_time)
exec_time.append(time.time() - start_time)
p3_utils.plot_line_graph(range(10, 101, 10), accuracy, "Knn Accuracy chart for faces",
"Percentage of training data used", "Accuracy obtained on test data")
p3_utils.plot_line_graph(range(10, 101, 10), exec_time, "Knn Runtime chart for faces",
"Percentage of training data used", "Run time in seconds")
print(statistics.stdev(accuracy))
print('errors over 3 iterations', errors)
print('Validating...')
validation_guesses = classifier.classify(validationDataList)
correct = [
validation_guesses[i] == face_data.face_validation_labels[i]
for i in range(len(face_data.face_validation_labels))
].count(True)
print(str(correct), 'correct out of ',
str(len(face_data.face_validation_labels)))
print('Testing...')
test_guesses = classifier.classify(testDataList)
correct = [
test_guesses[i] == face_data.face_test_labels[i]
for i in range(len(face_data.face_test_labels))
].count(True)
percentage = (100.0 * correct / len(face_data.face_test_labels))
print(str(correct), 'correct out of ',
str(len(face_data.face_test_labels)), 'percentage ', percentage)
percentages.append(percentage)
runtime = time.time() - start_time
print("Runtime taken: ", runtime)
runtimes.append(runtime)
p3_utils.plot_line_graph(range(10, 101, 10), percentages,
"Accuracy of perceptron for face data",
"Percentage of training data used",
"Percentage accuracy obtained on test data")
p3_utils.plot_line_graph(range(10, 101, 10), runtimes,
"Runtime of perceptron for face data",
"Percentage of training data used",
"Runtime taken for trianing and testing")
print(statistics.stdev(percentages))