点击查看 (人工智能) 系列文章

---

数据整理

样本数量统计

源码

# author: suoxd123@126.com
import tensorflow as tf# 读取数据
mnist = tf.keras.datasets.mnist
(imgTrain, labelTrain),(imgTest, labelTest) = mnist.load_data(path='mnist.npz')

# author: suoxd123@126.com
# 将2维矩阵变为1维向量
print('source data structure')
print(imgTrain.shape, type(imgTrain))
print(imgTest.shape, type(imgTest))imgTrain = imgTrain.reshape(60000, 784)
imgTest = imgTest.reshape(10000, 784)print('data structure after reshape')
print(imgTrain.shape, type(imgTrain))
print(imgTest.shape, type(imgTest))# 数据归一化处理
imgTrain = imgTrain.astype('float32')
imgTest = imgTest.astype('float32')
imgTrain /= 255
imgTest /= 255

# author: suoxd123@126.com
# 统计当前实际数据
import numpy as np
import matplotlib.pyplot as pltlabel, count = np.unique(labelTrain, return_counts=True)# 打印数值对应数量
print('0-9各数值对应的训练样本个数:')
for m,n in zip(label, count): print("%d:%d;" % (m, n), end='')
# 绘图显示对应数量
fig = plt.figure()
plt.bar(label, count, width = 0.6, align = 'center')
plt.title('0-9 label distruibution')
plt.xlabel('samble data')
plt.ylabel('quantity')
plt.xticks(label)
plt.ylim(0,8000) #设置y轴边界
for m,n in zip(label, count):plt.text(m,n, '%d'%n , ha = 'center', va = 'bottom', fontsize = 10)