def __init__(self,
train=True,
transform=None,
target_transform=None,
noise_rate=0.2,
split_percentage=0.9,
seed=1,
num_classes=10,
feature_size=28 * 28,
norm_std=0.1):
self.transform = transform
self.target_transform = target_transform
self.train = train
original_images = np.load('data/mnist/train_images.npy')
original_labels = np.load('data/mnist/train_labels.npy')
data = torch.from_numpy(original_images).float()
targets = torch.from_numpy(original_labels)
dataset = zip(data, targets)
new_labels = tools.get_instance_noisy_label(noise_rate, dataset,
targets, num_classes,
feature_size, norm_std,
seed)
self.train_data, self.val_data, self.train_labels, self.val_labels = tools.data_split(
original_images, new_labels, split_percentage, seed)
def __init__(self,
train=True,
transform=None,
target_transform=None,
noise_rate=0.2,
split_percentage=0.9,
seed=1,
num_classes=10,
feature_size=3 * 32 * 32,
norm_std=0.1):
self.transform = transform
self.target_transform = target_transform
self.train = train
original_images = np.load('data/svhn/train_images.npy')
original_labels = np.load('data/svhn/train_labels.npy')
data = torch.from_numpy(original_images).float()
targets = torch.from_numpy(original_labels)
dataset = zip(data, targets)
new_labels = tools.get_instance_noisy_label(noise_rate, dataset,
targets, num_classes,
feature_size, norm_std,
seed)
self.train_data, self.val_data, self.train_labels, self.val_labels = tools.data_split(
original_images, new_labels, split_percentage, seed)
if self.train:
self.train_data = self.train_data.reshape((-1, 3, 32, 32))
self.train_data = self.train_data.transpose((0, 2, 3, 1))
else:
self.val_data = self.val_data.reshape((-1, 3, 32, 32))
self.val_data = self.val_data.transpose((0, 2, 3, 1))
"""
print __doc__
import matplotlib.pyplot as plt
import numpy as np
import sys
sys.path.append("Kaggle Competetions/House Prices")
from tools import data_format
train_data = data_format("House Prices/House Prices - Data/train.csv")
test_data = data_format("House Prices/House Prices - Data/test.csv")
from tools import data_split
from sklearn.model_selection import train_test_split
targets, features, features_test_final = data_split(train_data, test_data)
features_train, features_test, labels_train, labels_test = train_test_split(
features, targets, test_size=0.9, random_state=42)
# Remove Outliers
import matplotlib.pyplot as plt
plt.subplot(1, 2, 1)
boxplot = plt.boxplot(labels_train, notch=True)
outliers = boxplot["fliers"][0].get_data()[1]
indices = [i for i, x in enumerate(labels_train) if x in outliers]
i = 0
for num in indices:
del labels_train[num - i]
del features_train[num - i]
i += 1