from DataInterface import get_pendigits_dataset, get_car_dataset, split_dataset
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.svm import LinearSVC
from sklearn.svm import SVC
from sklearn.metrics import accuracy_score
if __name__ == '__main__':
dataset = get_car_dataset()
data = split_dataset(dataset, 0.25)
train, test = data
features_test, labels_test = test
training_accuracy = []
testing_accuracy = []
for train_size in range(6, 99, 2):
train_size /= 100.0
data = split_dataset(train, 1 - train_size)
if train_size * 100 == 100.0:
print("Bro")
train2, test2 = data
features_train, labels_train = train2
rbf_kernel_svm_clf = SVC(kernel='rbf', gamma='auto', C=10)
rbf_kernel_svm_clf.fit(features_train, labels_train)
predictions = rbf_kernel_svm_clf.predict(features_train)
training_accuracy.append(accuracy_score(labels_train, predictions))
predictions = rbf_kernel_svm_clf.predict(features_test)
num_iter = 10
print(
"Now training and testing decision tree on car dataset for 10 runs of train/test split:\n"
)
features_name = [
"buying", "maint", "doors", "persons", "lug_boot", "safety"
]
dataset = get_car_dataset()
accuracies = 0
max_depth = 10
start_time = time.time()
print('To prune the tree, the maximum depth is set to ' + str(max_depth))
for _ in range(num_iter):
data = split_dataset(dataset, 0.25)
train, test = data
features_train, labels_train = train
dt = train_decision_tree(data, max_depth)
accuracy = test_decision_tree(data, dt)
accuracies += accuracy
duration = time.time() - start_time
print("Average accuracy is {0:.3f}.\n".format(accuracies / num_iter))
print("The run time is " + str(duration) + " sec.")
print("\nComplete.\n")
print(
"-----------------------------------------------------------------\n")
print(
"Now training and testing decision tree on pen digits dataset for 10 runs of train/test split:\n"
if __name__ == '__main__':
num_iter = 10 # number of test/train splits
print("\nNow training and testing on the car dataset with " + str(num_iter) + " runs of train/test splits:\n")
data = get_car_dataset()
hidden_layers = (10,10,10,10)
print("The neural network has {} hidden layers, each layer has size: ".format(len(hidden_layers))),
for layer in hidden_layers:
print(layer),
print('\n')
accuracies = 0
start_time = time.time()
for _ in range(num_iter):
train, test = split_dataset(data, 0.25)
mlp = training(train, hidden_layers)
accuracy = testing(mlp, test)
accuracies += accuracy
duration = time.time() - start_time
print("The average classification rate is {0:.3f}.\n".format(accuracies / num_iter))
print("The run time is " + str(duration) + " sec.")
print("\nComplete\n----------------------------------\n")
print("Now training and testing on the pen digits dataset " + str(num_iter) + " runs of train/test splits:\n")
data = get_pendigits_dataset()
hidden_layers = (10,10)
print("The neural network has {} hidden layers, each layer has size: ".format(len(hidden_layers))),
for layer in hidden_layers:
print(layer),
print(
"\nNow training and testing boosted version of decision tree on car dataset with train/test split for 10 times:\n"
)
dataset = get_car_dataset()
feature_names = [
"buying", "maint", "doors", "persons", "lug_boot", "safety"
]
max_depth = 4
print("To prune the decision tree, the maximum depth is set to " +
str(max_depth))
start_time = time.time()
accuracies = 0
for _ in range(10):
data = split_dataset(dataset, 0.3)
train, test = data
features_train, labels_train = train
clf = train_boosting(data, max_depth)
accuracy = test_boosting(data, clf)
accuracies += accuracy
# save_trees_as_png(clf, feature_names, 'car')
# print("Trees visualization written to the current folder.")
duration = time.time() - start_time
print("The average training accuracy over 10 runs is {0:.3f}.\n".format(
accuracies / 10))
print("The run time is " + str(duration) + " sec.")
print("\nComplete.\n")
print(
"-----------------------------------------------------------------\n")