The objective of this assignment is to learn about simple data curation practices, and familiarize you with some of the data we'll be reusing later.
This notebook uses the notMNIST dataset to be used with python experiments. This dataset is designed to look like the classic MNIST dataset, while looking a little more like real data: it's a harder task, and the data is a lot less 'clean' than MNIST.
In [1]:
# These are all the modules we'll be using later. Make sure you can import them
# before proceeding further.
from __future__ import print_function
import matplotlib.pyplot as plt
import numpy as np
import os
import sys
import tarfile
from IPython.display import display, Image
from scipy import ndimage
from sklearn.linear_model import LogisticRegression
from six.moves.urllib.request import urlretrieve
from six.moves import cPickle as pickle
# Config the matplotlib backend as plotting inline in IPython
%matplotlib inline
First, we'll download the dataset to our local machine. The data consists of characters rendered in a variety of fonts on a 28x28 image. The labels are limited to 'A' through 'J' (10 classes). The training set has about 500k and the testset 19000 labelled examples. Given these sizes, it should be possible to train models quickly on any machine.
In [2]:
url = 'http://commondatastorage.googleapis.com/books1000/'
last_percent_reported = None
data_root = '.' # Change me to store data elsewhere
def download_progress_hook(count, blockSize, totalSize):
"""A hook to report the progress of a download. This is mostly intended for users with
slow internet connections. Reports every 5% change in download progress.
"""
global last_percent_reported
percent = int(count * blockSize * 100 / totalSize)
if last_percent_reported != percent:
if percent % 5 == 0:
sys.stdout.write("%s%%" % percent)
sys.stdout.flush()
else:
sys.stdout.write(".")
sys.stdout.flush()
last_percent_reported = percent
def maybe_download(filename, expected_bytes, force=False):
"""Download a file if not present, and make sure it's the right size."""
dest_filename = os.path.join(data_root, filename)
if force or not os.path.exists(dest_filename):
print('Attempting to download:', filename)
filename, _ = urlretrieve(url + filename, dest_filename, reporthook=download_progress_hook)
print('\nDownload Complete!')
statinfo = os.stat(dest_filename)
if statinfo.st_size == expected_bytes:
print('Found and verified', dest_filename)
else:
raise Exception(
'Failed to verify ' + dest_filename + '. Can you get to it with a browser?')
return dest_filename
train_filename = maybe_download('notMNIST_large.tar.gz', 247336696)
test_filename = maybe_download('notMNIST_small.tar.gz', 8458043)
Extract the dataset from the compressed .tar.gz file. This should give you a set of directories, labelled A through J.
In [3]:
num_classes = 10
np.random.seed(133)
def maybe_extract(filename, force=False):
root = os.path.splitext(os.path.splitext(filename)[0])[0] # remove .tar.gz
if os.path.isdir(root) and not force:
# You may override by setting force=True.
print('%s already present - Skipping extraction of %s.' % (root, filename))
else:
print('Extracting data for %s. This may take a while. Please wait.' % root)
tar = tarfile.open(filename)
sys.stdout.flush()
tar.extractall(data_root)
tar.close()
data_folders = [
os.path.join(root, d) for d in sorted(os.listdir(root))
if os.path.isdir(os.path.join(root, d))]
if len(data_folders) != num_classes:
raise Exception(
'Expected %d folders, one per class. Found %d instead.' % (
num_classes, len(data_folders)))
print(data_folders)
return data_folders
train_folders = maybe_extract(train_filename)
test_folders = maybe_extract(test_filename)
In [4]:
def display_random_image(folder, num_to_display):
""" Display a number of images from a folder. """
files = os.listdir(folder)
files_sample = np.random.choice(files, num_to_display)
for file_sample in files_sample:
display(Image(filename=os.path.join(folder, file_sample)))
for folder in train_folders:
display_random_image(folder, 5)
for folder in test_folders:
display_random_image(folder, 5)
Now let's load the data in a more manageable format. Since, depending on your computer setup you might not be able to fit it all in memory, we'll load each class into a separate dataset, store them on disk and curate them independently. Later we'll merge them into a single dataset of manageable size.
We'll convert the entire dataset into a 3D array (image index, x, y) of floating point values, normalized to have approximately zero mean and standard deviation ~0.5 to make training easier down the road.
A few images might not be readable, we'll just skip them.
In [5]:
image_size = 28 # Pixel width and height.
pixel_depth = 255.0 # Number of levels per pixel.
def load_letter(folder, min_num_images):
"""Load the data for a single letter label."""
image_files = os.listdir(folder)
dataset = np.ndarray(shape=(len(image_files), image_size, image_size),
dtype=np.float32)
print(folder)
num_images = 0
for image in image_files:
image_file = os.path.join(folder, image)
try:
image_data = (ndimage.imread(image_file).astype(float) -
pixel_depth / 2) / pixel_depth
if image_data.shape != (image_size, image_size):
raise Exception('Unexpected image shape: %s' % str(image_data.shape))
dataset[num_images, :, :] = image_data
num_images = num_images + 1
except IOError as e:
print('Could not read:', image_file, ':', e, '- it\'s ok, skipping.')
dataset = dataset[0:num_images, :, :]
if num_images < min_num_images:
raise Exception('Many fewer images than expected: %d < %d' %
(num_images, min_num_images))
print('Full dataset tensor:', dataset.shape)
print('Mean:', np.mean(dataset))
print('Standard deviation:', np.std(dataset))
return dataset
def maybe_pickle(data_folders, min_num_images_per_class, force=False):
dataset_names = []
for folder in data_folders:
set_filename = folder + '.pickle'
dataset_names.append(set_filename)
if os.path.exists(set_filename) and not force:
# You may override by setting force=True.
print('%s already present - Skipping pickling.' % set_filename)
else:
print('Pickling %s.' % set_filename)
dataset = load_letter(folder, min_num_images_per_class)
try:
with open(set_filename, 'wb') as f:
pickle.dump(dataset, f, pickle.HIGHEST_PROTOCOL)
except Exception as e:
print('Unable to save data to', set_filename, ':', e)
return dataset_names
train_datasets = maybe_pickle(train_folders, 45000)
test_datasets = maybe_pickle(test_folders, 1800)
In [6]:
def display_random_matrix(pickle_file, num_to_display):
""" Display a number of images from a folder. """
with open(pickle_file, 'rb') as f:
dataset = pickle.load(f)
index_samples = np.random.choice(range(0, dataset.shape[0]), num_to_display)
for index_sample in index_samples:
plt.figure()
plt.imshow(dataset[index_sample, :, :])
for pickle_file in train_datasets:
display_random_matrix(pickle_file, 3)
for pickle_file in test_datasets:
display_random_matrix(pickle_file, 1)
In [7]:
def display_class_size(pickle_file):
""" Number of samples in each class. """
with open(pickle_file, 'rb') as f:
dataset = pickle.load(f)
print("Dataset {0} is of size: {1}".format(pickle_file, dataset.shape[0]))
for pickle_file in train_datasets:
display_class_size(pickle_file)
for pickle_file in test_datasets:
display_class_size(pickle_file)
Merge and prune the training data as needed. Depending on your computer setup, you might not be able to fit it all in memory, and you can tune train_size as needed. The labels will be stored into a separate array of integers 0 through 9.
Also create a validation dataset for hyperparameter tuning.
In [8]:
def make_arrays(nb_rows, img_size):
if nb_rows:
dataset = np.ndarray((nb_rows, img_size, img_size), dtype=np.float32)
labels = np.ndarray(nb_rows, dtype=np.int32)
else:
dataset, labels = None, None
return dataset, labels
def merge_datasets(pickle_files, train_size, valid_size=0):
num_classes = len(pickle_files)
valid_dataset, valid_labels = make_arrays(valid_size, image_size)
train_dataset, train_labels = make_arrays(train_size, image_size)
vsize_per_class = valid_size // num_classes
tsize_per_class = train_size // num_classes
start_v, start_t = 0, 0
end_v, end_t = vsize_per_class, tsize_per_class
end_l = vsize_per_class+tsize_per_class
for label, pickle_file in enumerate(pickle_files):
try:
with open(pickle_file, 'rb') as f:
letter_set = pickle.load(f)
# let's shuffle the letters to have random validation and training set
np.random.shuffle(letter_set)
if valid_dataset is not None:
valid_letter = letter_set[:vsize_per_class, :, :]
valid_dataset[start_v:end_v, :, :] = valid_letter
valid_labels[start_v:end_v] = label
start_v += vsize_per_class
end_v += vsize_per_class
train_letter = letter_set[vsize_per_class:end_l, :, :]
train_dataset[start_t:end_t, :, :] = train_letter
train_labels[start_t:end_t] = label
start_t += tsize_per_class
end_t += tsize_per_class
except Exception as e:
print('Unable to process data from', pickle_file, ':', e)
raise
return valid_dataset, valid_labels, train_dataset, train_labels
train_size = 200000
valid_size = 10000
test_size = 10000
valid_dataset, valid_labels, train_dataset, train_labels = merge_datasets(
train_datasets, train_size, valid_size)
_, _, test_dataset, test_labels = merge_datasets(test_datasets, test_size)
print('Training:', train_dataset.shape, train_labels.shape)
print('Validation:', valid_dataset.shape, valid_labels.shape)
print('Testing:', test_dataset.shape, test_labels.shape)
Next, we'll randomize the data. It's important to have the labels well shuffled for the training and test distributions to match.
In [9]:
def randomize(dataset, labels):
permutation = np.random.permutation(labels.shape[0])
shuffled_dataset = dataset[permutation,:,:]
shuffled_labels = labels[permutation]
return shuffled_dataset, shuffled_labels
train_dataset, train_labels = randomize(train_dataset, train_labels)
test_dataset, test_labels = randomize(test_dataset, test_labels)
valid_dataset, valid_labels = randomize(valid_dataset, valid_labels)
In [10]:
def show_random_samples(dataset, labels, num_to_display):
index_samples = np.random.choice(range(0, dataset.shape[0]), num_to_display)
for index_sample in index_samples:
print(labels[index_sample])
plt.matshow(dataset[index_sample, :, :])
show_random_samples(train_dataset, train_labels, 5)
show_random_samples(test_dataset, test_labels, 3)
show_random_samples(valid_dataset, valid_labels, 3)
Finally, let's save the data for later reuse:
In [12]:
pickle_file = os.path.join(data_root, 'notMNIST.pickle')
try:
f = open(pickle_file, 'wb')
save = {
'train_dataset': train_dataset,
'train_labels': train_labels,
'valid_dataset': valid_dataset,
'valid_labels': valid_labels,
'test_dataset': test_dataset,
'test_labels': test_labels,
}
pickle.dump(save, f, pickle.HIGHEST_PROTOCOL)
f.close()
except Exception as e:
print('Unable to save data to', pickle_file, ':', e)
raise
In [13]:
statinfo = os.stat(pickle_file)
print('Compressed pickle size:', statinfo.st_size)
By construction, this dataset might contain a lot of overlapping samples, including training data that's also contained in the validation and test set! Overlap between training and test can skew the results if you expect to use your model in an environment where there is never an overlap, but are actually ok if you expect to see training samples recur when you use it. Measure how much overlap there is between training, validation and test samples.
Optional questions:
In [40]:
train_dataset[455, :, :].shape
Out[40]:
In [43]:
overlap_train_validation = 0
overlap_train_test = 0
overlap_validation_test = 0
for train_index in range(0, train_dataset.shape[0]):
image_train = train_dataset[train_index, :, :]
for valid_index in range(0, valid_dataset.shape[0]):
image_valid = valid_dataset[valid_index, :, :]
if np.array_equal(image_train, image_valid):
overlap_train_validation += 1
for test_index in range(0, test_dataset.shape[0]):
image_test = test_dataset[test_index, :, :]
if np.array_equal(image_train, image_test):
overlap_train_test += 1
for valid_index in range(0, valid_dataset.shape[0]):
image_valid = valid_dataset[valid_index, :, :]
for test_index in range(0, test_dataset.shape[0]):
image_test = test_dataset[test_index, :, :]
if np.array_equal(image_valid, image_test):
overlap_validation_test += 1
print("Overlap between training and validation datasets = {0}".format(overlap_train_validation))
print("Overlap between training and test datasets = {0}".format(overlap_train_test))
print("Overlap between test and validation datasets = {0}".format(overlap_validation_test))
Let's get an idea of what an off-the-shelf classifier can give you on this data. It's always good to check that there is something to learn, and that it's a problem that is not so trivial that a canned solution solves it.
Train a simple model on this data using 50, 100, 1000 and 5000 training samples. Hint: you can use the LogisticRegression model from sklearn.linear_model.
Optional question: train an off-the-shelf model on all the data!
In [ ]:
def model_test(train_dataset, train_labels, test_dataset, test_labels, number_of_samples=None):
if not number_of_samples or number_of_samples > train_dataset.shape[0]:
samples = range(0, train_dataset.shape[0])
else:
samples = np.random.choice(range(0, train_dataset.shape[0]), number_of_samples)
logreg = LogisticRegression()
training_data = train_dataset[samples, :, :]
training_labels = train_labels[samples]
X = np.ndarray((training_data.shape[0], training_data.shape[1] * training_data.shape[2]), dtype=np.float32)
Z = np.ndarray((test_dataset.shape[0], test_dataset.shape[1] * test_dataset.shape[2]), dtype=np.float32)
# Need to convert data to 2D rather than 3D array - convert image into 1D
for train_index in range(0, training_data.shape[0]):
X[train_index] = np.ndarray.flatten(training_data[train_index, :, :])
for test_index in range(0, test_dataset.shape[0]):
Z[test_index] = np.ndarray.flatten(test_dataset[test_index, :, :])
logreg.fit(X, training_labels)
predicted = logreg.predict(Z)
print("With {0} training samples, accuracy is {1}".format(number_of_samples, np.mean(predicted == test_labels)*100))
model_test(train_dataset, train_labels, test_dataset, test_labels, number_of_samples=50)
model_test(train_dataset, train_labels, test_dataset, test_labels, number_of_samples=100)
model_test(train_dataset, train_labels, test_dataset, test_labels, number_of_samples=1000)
model_test(train_dataset, train_labels, test_dataset, test_labels, number_of_samples=5000)
model_test(train_dataset, train_labels, test_dataset, test_labels)
Can use "score" function in scikit learn to test using testX and testY.