def main():
data_file = 'seeds_dataset_shuffled.txt'
instances, labels = read_data(data_file)
# want labels run from 0 through K-1
# not 1 through K
labels = [i-1 for i in labels]
print 'Read %d instances and %d labels from file %s.' \
% (len(instances), len(labels), data_file)
if len(instances) != len(labels):
raise Exception('Expected equal number of instances and labels.')
else:
n = len(instances)
# Find number of clusters by finding out how many unique elements are there
# in labels.
K = num_unique_labels(labels)
print 'Found %d unique labels.' % K
# k-nearest neighbor classification for various k
k_range = range(1, 31)
# create empty list to store cross-validation errors for different k
cv_error = []
# 10-fold cross-validation
num_folds = 10
for k in k_range:
total_error = 0.0
# create and process all folds
for fold_id in range(num_folds):
# separate indices into training indices
# and test indices for this particular fold
fold_train_indices, fold_test_indices = \
get_fold_indices(n, num_folds, fold_id)
# TASK 3.4.1
# get the training data and labels
# and create a k-NN classifier
train_data = [instances[x] for x in fold_train_indices]
train_labels = [labels[x] for x in fold_train_indices]
classifier = lambda p: \
nn_classifier(p, train_data, train_labels, k, K)
# TASK 3.4.2
# get the test data and labels
# and evaluate the classifier's error
test_data = [instances[x] for x in fold_test_indices]
test_label = [labels[x] for x in fold_test_indices]
fold_error = classification_error(classifier,
test_data, test_label)
total_error += fold_error
cv_error.append(total_error/num_folds)
# print the values for k and the corresponding cross-validation errors
for i in range(len(k_range)):
print k_range[i], cv_error[i]
def main():
data_file = 'seeds_dataset_shuffled.txt'
instances, labels = read_data(data_file)
# want labels run from 0 through K-1
# not 1 through K
labels = [i - 1 for i in labels]
print 'Read %d instances and %d labels from file %s.' \
% (len(instances), len(labels), data_file)
if len(instances) != len(labels):
raise Exception('Expected equal number of instances and labels.')
else:
n = len(instances)
# Find number of clusters by finding out how many unique elements are there
# in labels.
K = num_unique_labels(labels)
print 'Found %d unique labels.' % K
# k-nearest neighbor classification for various k
k_range = range(1, 31)
# create empty list to store cross-validation errors for different k
cv_error = []
# 10-fold cross-validation
num_folds = 10
for k in k_range:
total_error = 0.0
# create and process all folds
for fold_id in range(num_folds):
# separate indices into training indices
# and test indices for this particular fold
fold_train_indices, fold_test_indices = \
get_fold_indices(n, num_folds, fold_id)
# get the training data and labels
# and create a k-NN classifier
train_data = [instances[i] for i in fold_train_indices]
train_labels = [labels[i] for i in fold_train_indices]
classifier = lambda x: nn_classifier(x, train_data, train_labels,
k, K)
# get the test data and labels
# and evaluate the classifier's error
test_data = [instances[i] for i in fold_test_indices]
test_label = [labels[i] for i in fold_test_indices]
fold_error = classification_error(classifier, test_data,
test_label)
total_error += fold_error
cv_error.append(total_error / num_folds)
# print the values for k and the corresponding cross-validation errors
for i in range(len(k_range)):
print k_range[i], cv_error[i]