MNIST-NN-Pure-Math/data.py at main · SawyerAlston/MNIST-NN-Pure-Math · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
import numpy as np
import pandas as pd
import os
import urllib.request

train_url = "https://python-course.eu/data/mnist/mnist_train.csv"
test_url = "https://python-course.eu/data/mnist/mnist_test.csv"
train_path = "data/mnist_train.csv"
test_path = "data/mnist_test.csv"

def download_data():
  os.makedirs("data", exist_ok=True)

  if not os.path.exists(train_path):
    print("grabbing train data...")
    urllib.request.urlretrieve(train_url, train_path)

  if not os.path.exists(test_path):
    print("grabbing test data...")
    urllib.request.urlretrieve(test_url, test_path)

def load_data():
  download_data()

  train_data = pd.read_csv(train_path).to_numpy()
  np.random.shuffle(train_data)
  train_X = train_data[:, 1:].T / 255.0
  train_y = train_data[:, 0]

  test_data = pd.read_csv(test_path).to_numpy()
  np.random.shuffle(test_data)
  test_X = test_data[:, 1:].T / 255.0
  test_y = test_data[:, 0]

  return train_X, train_y, test_X, test_y