NumPy（六）：数组堆叠：【vstack：垂直（按列顺序）堆叠数组】【hstack：水平（按列顺序）堆叠数组】【stack：axis=0/1/2】

ゝ一纸荒年。 2023-09-29 08:39 10阅读 0赞

首先生成一些数，

import numpy as np
    a = np.arange(1, 7).reshape((2, 3))
    b = np.arange(7, 13).reshape((2, 3))
    c = np.arange(13, 19).reshape((2, 3))
    
    print('a = \n', a)
    print('b = \n', b)
    print('c = \n', c)

即下面的形式  
![在这里插入图片描述][20201209181849872.png_pic_center]  
下面分别以不同的形式输出：

## 一、vstack ##

s = np.vstack((a, b, c))
    print('vstack \n ', s.shape, '\n', s)

在竖直方向上进行堆叠,得到6×3的矩阵

![在这里插入图片描述][20201209181931274.png_pic_center]

**vstack(tup)：垂直（按列顺序）堆叠数组：**

import numpy as np
    
    # 数组堆叠
    
    # numpy.vstack(tup)：垂直（按列顺序）堆叠数组
    a = np.arange(5)
    b = np.arange(5, 10)
    ar1 = np.vstack((a, b))
    print('a = {0}, a.shape = {1}'.format(a, a.shape))
    print('b = {0}, b.shape = {1}'.format(b, b.shape))
    print('ar1 = {0}, ar1.shape = {1}'.format(ar1, ar1.shape))
    a = np.array([[1], [2], [3]])
    b = np.array([['a'], ['b'], ['c'], ['d']])
    ar2 = np.vstack((a, b))  # 这里形状可以不一样
    print('a = {0}, a.shape = {1}'.format(a, a.shape))
    print('b = {0}, b.shape = {1}'.format(b, b.shape))
    print('ar2 = {0}, ar2.shape = {1}'.format(ar2, ar2.shape))

打印结果：

a = [0 1 2 3 4], a.shape = (5,)
    b = [5 6 7 8 9], b.shape = (5,)
    ar1 = [[0 1 2 3 4]
     [5 6 7 8 9]], ar1.shape = (2, 5)
    --------------------------------------------------
    a = [[1]
     [2]
     [3]], a.shape = (3, 1)
    b = [['a']
     ['b']
     ['c']
     ['d']], b.shape = (4, 1)
    ar2 = [['1']
     ['2']
     ['3']
     ['a']
     ['b']
     ['c']
     ['d']], ar2.shape = (7, 1)

## 二、hstack ##

s = np.hstack((a, b, c))
    print('hstack \n ', s.shape, '\n', s)

![在这里插入图片描述][20201209182003411.png_pic_center]

**hstack(tup)：水平（按列顺序）堆叠数组：**

import numpy as np
    
    # 数组堆叠
    
    # numpy.hstack(tup)：水平（按列顺序）堆叠数组
    a = np.arange(5)  # a为一维数组，5个元素
    b = np.arange(5, 9)  # b为一维数组,4个元素
    ar1 = np.hstack((a, b))  # 注意:((a,b))，这里形状可以不一样
    print('a = {0}, a.shape = {1}'.format(a, a.shape))
    print('b = {0}, b.shape = {1}'.format(b, b.shape))
    print('ar1 = {0}, ar1.shape = {1}'.format(ar1, ar1.shape))
    print('-' * 50)
    a = np.array([[1], [2], [3]])  # a为二维数组，3行1列
    b = np.array([['a'], ['b'], ['c']])  # b为二维数组，3行1列
    ar2 = np.hstack((a, b))  # 注意:((a,b))，这里形状必须一样
    print('a = {0}, a.shape = {1}'.format(a, a.shape))
    print('b = {0}, b.shape = {1}'.format(b, b.shape))
    print('ar2 = {0}, ar2.shape = {1}'.format(ar2, ar2.shape))

打印结果：

a = [0 1 2 3 4], a.shape = (5,)
    b = [5 6 7 8], b.shape = (4,)
    ar1 = [0 1 2 3 4 5 6 7 8], ar1.shape = (9,)
    --------------------------------------------------
    a = [[1]
     [2]
     [3]], a.shape = (3, 1)
    b = [['a']
     ['b']
     ['c']], b.shape = (3, 1)
    ar2 = [['1' 'a']
     ['2' 'b']
     ['3' 'c']], ar2.shape = (3, 2)

## 三、stack ##

### 1、axis=0 ###

![在这里插入图片描述][20201209181849872.png_pic_center]

s = np.stack((a, b, c), axis=0)
    print('axis = 0 \n ', s.shape, '\n', s)

就是下面的形式

![在这里插入图片描述][20201209182039412.png_pic_center]  
具体为3组2×3矩阵  
![在这里插入图片描述][20201209182047419.png_pic_center]

stack(arrays, axis=0)：沿着新轴连接数组的序列，形状必须一样！  
重点解释axis参数的意思，假设两个数组\[1 2 3\]和\[4 5 6\]，shape均为(3,0)

*  axis=0：\[\[1 2 3\] \[4 5 6\]\]，shape为(2,3)
 *  axis=1：\[\[1 4\] \[2 5\] \[3 6\]\]，shape为(3,2)

import numpy as np
    
    # 数组堆叠
    # numpy.stack(arrays, axis=0)：沿着新轴连接数组的序列，形状必须一样！
    # 重点解释axis参数的意思，假设两个数组[1 2 3]和[4 5 6]，shape均为(3,0)
    # axis=0：[[1 2 3] [4 5 6]]，shape为(2,3)
    # axis=1：[[1 4] [2 5] [3 6]]，shape为(3,2)
    a = np.arange(5)
    b = np.arange(5, 10)
    ar1 = np.stack((a, b))
    ar2 = np.stack((a, b), axis=1)
    print('a = {0}, a.shape = {1}'.format(a, a.shape))
    print('b = {0}, b.shape = {1}'.format(b, b.shape))
    print('ar1 = {0}, ar1.shape = {1}'.format(ar1, ar1.shape))
    print('ar2 = {0}, ar2.shape = {1}'.format(ar2, ar2.shape))

打印结果：

a = [0 1 2 3 4], a.shape = (5,)
    b = [5 6 7 8 9], b.shape = (5,)
    ar1 = [[0 1 2 3 4]
     [5 6 7 8 9]], ar1.shape = (2, 5)
    ar2 = [[0 5]
     [1 6]
     [2 7]
     [3 8]
     [4 9]], ar2.shape = (5, 2)

### 2、axis=1 ###

![在这里插入图片描述][20201209181849872.png_pic_center]

s = np.stack((a, b, c), axis=1)
    print('axis = 1 \n ', s.shape, '\n', s)

即将每个矩阵的每一行进行堆叠，放在一个矩阵里（一行对应一个矩阵）就是下图的红色的放一起，绿色的放一起

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTMyNTA4NjE_size_16_color_FFFFFF_t_70_pic_center]

### 3、axis=2 ###

![在这里插入图片描述][20201209181849872.png_pic_center]

s = np.stack((a, b, c), axis=2)
    print('axis = 2 \n ', s.shape, '\n', s)

即将每行的进行竖排，放在一个矩阵里（一行对应一个矩阵）  
![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTMyNTA4NjE_size_16_color_FFFFFF_t_70_pic_center 1]

[20201209181849872.png_pic_center]: https://img-blog.csdnimg.cn/20201209181849872.png#pic_center
[20201209181931274.png_pic_center]: https://img-blog.csdnimg.cn/20201209181931274.png#pic_center
[20201209182003411.png_pic_center]: https://img-blog.csdnimg.cn/20201209182003411.png#pic_center
[20201209182039412.png_pic_center]: https://img-blog.csdnimg.cn/20201209182039412.png#pic_center
[20201209182047419.png_pic_center]: https://img-blog.csdnimg.cn/20201209182047419.png#pic_center
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTMyNTA4NjE_size_16_color_FFFFFF_t_70_pic_center]: https://img-blog.csdnimg.cn/20201209182115857.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTMyNTA4NjE=,size_16,color_FFFFFF,t_70#pic_center
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTMyNTA4NjE_size_16_color_FFFFFF_t_70_pic_center 1]: https://img-blog.csdnimg.cn/20201209182138507.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3UwMTMyNTA4NjE=,size_16,color_FFFFFF,t_70#pic_center