1.数据集准备:
import sklearn
assert sklearn.__version__ >= "0.20"
import numpy as np
import os
from sklearn.datasets import fetch_openml
mnist = fetch_openml('mnist_784', version=1, as_frame=False)
mnist.keys()
输出:
dict_keys(['data', 'target', 'frame', 'categories', 'feature_names', 'target_names', 'DESCR', 'details', 'url'])
如果网络信号不好,可加载下载好的数据集,速度更快:
from scipy.io import loadmat
mnist = loadmat('/app/datasets/mnist-original.mat')
mnist.keys()
2.数据集结构为:
X, y = mnist["data"], mnist["label"]
#交换两个维度
X=np.swapaxes(X,0,1)
print(X.shape)
y=np.squeeze(y)
print(y.shape)
输出:
(70000, 784)
(70000,)
3.拼合并展示数据集:
def plot_digits(instances, images_per_row=10, **options):
size = 28
images_per_row = min(len(instances), images_per_row)
# 计算行数
n_rows = (len(instances) - 1) // images_per_row + 1
# 用空白补齐缺少的图片:
n_empty = n_rows * images_per_row - len(instances)
padded_instances = np.concatenate([instances, np.zeros((n_empty, size * size))], axis=0)
# 重塑形状 28×28 :
image_grid = padded_instances.reshape((n_rows, images_per_row, size, size))
#使用transpose以便于reshape相邻元素:
big_image = image_grid.transpose(0, 2, 1, 3).reshape(n_rows * size, images_per_row * size)
plt.imshow(big_image, cmap = mpl.cm.binary, **options)
plt.axis("off")
import matplotlib as mpl
import matplotlib.pyplot as plt
plt.figure(figsize=(9,9))
example_images = X[:100]
plot_digits(example_images, images_per_row=10)
plt.show()