'''
# Load data using sklearn
split_point = 100
digits = dataset.load_digits(2)
features = digits['data']
label = digits['target']
label[label == 1] = 1
label[label == 0] = -1
# Display a sample char
display_char(digits, 10)
# Standardize Dataset
features = preprocessing.scale(features)
# split train and test
train_features = features[0:split_point, :]
train_label = label[0:split_point]
test_features = features[split_point + 1:, :]
test_label = label[split_point + 1:]
_, n_feat = features.shape
# Train SVM
my_svm = SVM.SVM_Pegasos(n_feat, 10000, 1.0)
my_svm.train(train_features, train_label)
# Test it
acc, result = my_svm.test(test_features, test_label, True)
print('Accuracy on test set is %1.4f' % acc)
