#-------------------------------------------------------------------------------
# Part 1 - Perceptron Algorithm
#-------------------------------------------------------------------------------
# toy_features, toy_labels = utils.load_toy_data('../../Data/toy_data.csv')
#
# theta, theta_0 = lab2.perceptron(toy_features, toy_labels, T=5)
#
# utils.plot_toy_results(toy_features, toy_labels, theta, theta_0)
#-------------------------------------------------------------------------------
# Part 2 - Classifying Reviews
#-------------------------------------------------------------------------------
theta, theta_0 = lab2.perceptron(train_bow_features, train_labels, T=5)
train_accuracy = lab2.accuracy(train_bow_features, train_labels, theta, theta_0)
val_accuracy = lab2.accuracy(val_bow_features, val_labels, theta, theta_0)
#
# print("Training accuracy: {:.4f}".format(train_accuracy))
# print("Validation accuracy: {:.4f}".format(val_accuracy))
#-------------------------------------------------------------------------------
# Part 3 - Improving the Model
#-------------------------------------------------------------------------------
#-------------------------------------------------------------------------------
# Part 3.1 - Tuning the Hyperparameters
#-------------------------------------------------------------------------------
bigram_dictionary = lab2.bigram_dictionary(train_texts)
train_final_features = lab2.extract_final_features(train_texts, dictionary,
bigram_dictionary)
val_final_features = lab2.extract_final_features(val_texts, dictionary,
bigram_dictionary)
test_final_features = lab2.extract_final_features(test_texts, dictionary,
bigram_dictionary)
#-------------------------------------------------------------------------------
# Part 1 - Perceptron Algorithm
#-------------------------------------------------------------------------------
toy_features, toy_labels = utils.load_toy_data('../../Data/toy_data.csv')
theta, theta_0 = lab2.perceptron(toy_features, toy_labels, T=5)
utils.plot_toy_results(toy_features, toy_labels, theta, theta_0)
#-------------------------------------------------------------------------------
# Part 2 - Classifying Reviews
#-------------------------------------------------------------------------------
theta, theta_0 = lab2.perceptron(train_bow_features, train_labels, T=5)
train_accuracy = lab2.accuracy(train_bow_features, train_labels, theta,
theta_0)
val_accuracy = lab2.accuracy(val_bow_features, val_labels, theta, theta_0)
print("Training accuracy: {:.4f}".format(train_accuracy)) # 0.9850
print("Validation accuracy: {:.4f}".format(val_accuracy)) # 0.8943
val_bow_features = lab2.extract_bow_feature_vectors(val_texts, dictionary)
test_bow_features = lab2.extract_bow_feature_vectors(test_texts, dictionary)
# You may modify the following when adding additional features (Part 3c)
train_final_features = lab2.extract_final_features(train_texts, dictionary)
val_final_features = lab2.extract_final_features(val_texts, dictionary)
test_final_features = lab2.extract_final_features(test_texts, dictionary)
#-------------------------------------------------------------------------------
# Part 1 - Perceptron Algorithm
#-------------------------------------------------------------------------------
toy_features, toy_labels = utils.load_toy_data('../../Data/toy_data.csv')
theta, theta_0 = lab2.perceptron(toy_features, toy_labels, T=5)
utils.plot_toy_results(toy_features, toy_labels, theta, theta_0)
#-------------------------------------------------------------------------------
# Part 2 - Classifying Reviews
#-------------------------------------------------------------------------------
#theta,theta_0 = lab2.perceptron(train_bow_features, train_labels, T=5)
#train_accuracy = lab2.accuracy(train_bow_features, train_labels, theta, theta_0)
#val_accuracy = lab2.accuracy(val_bow_features, val_labels, theta, theta_0)
#print("Training accuracy: {:.4f}".format(train_accuracy))
#print("Validation accuracy: {:.4f}".format(val_accuracy))