焦海洋的Numpy作业

垂直拆分：numpy.vsplit(数组，份数)->(数组片段)
import numpy as np
c = np.arange(1,13).reshape(6,2)
c
array([1,2],[3,4],[5,6],[7,8],[9,10],[11,12])
np.vsplit(c,3)
[array([[1,2],[3,4],])array([[5,6],[7,8]])array([[9,10],[11,12]])]


水平拆分：numpy.hsplit(数组，份数)->(数组片段)
d =c.T
d
array([[1,3,5,7,9,11],[2,4,6,8,10,12]])
array([[1, 3],[2, 4]]), array([[5, 7],[6, 8]]), array([[ 9, 11],[10, 12]])

numpy.dsplit(数组，份数)->(数组片段)
import numpy as np
e = np.dstack((a, b))
e
array([[[11, 21],[12, 22],[13, 23]],[[14, 24],[15, 25],[16, 26]],[[17, 27],[18, 28],[19, 29]]])
np.dsplit(e, 2)
Out[35]:
[array([[[11],[12],[13]],[[14],[15],[16]],[[17],[18],[19]]]),
 array([[[21],[22],[23]],[[24],[25],[26]],[[27],[28],[29]]])]

灵活选择引用
inistate =np.array([1,2,3,4])
pre_inistate = inistate[0:3]
pre_inistate
array([0.18257419, 0.36514837, 0.54772256])

numpy基本加减和取行操作
import numpy as np
a = np.array([1,1,1,1])
b = np.array([[1],[1],[1],[1]])
a+b
array([[2, 2, 2, 2],
[2, 2, 2, 2],
[2, 2, 2, 2],
[2, 2, 2, 2]])

c = np.array([[1,1,1,1]])
c+b
array([[2, 2, 2, 2],
[2, 2, 2, 2],
[2, 2, 2, 2],
[2, 2, 2, 2]])

W = np.array([[1,1,1],[2,2,2]])
W[:,1]
array([1, 2])

W[1]
array([2, 2, 2])

W[:,1] = np.array([5,5])
W
array([[1, 5, 1],
[2, 5, 2]])

delete()函数
import numpy as np
matrix = [[1,2,3,4],[5,6,7,8],[9,10,11,12]]
p1 = np.delete(matrix, 1, 0) # 第0维度(行)第1行被删除(初始行为0行)
print(’>>>>p1>>>>\n’,p1)
p2 = np.delete(matrix, 1, 1) # 第1维度(列)第1行被删除
print(’>>>>p2>>>>\n’,p2)
p3 = np.delete(matrix, 1) # 拉平后删除第1个元素(初始为第0个)
print(’>>>>p3>>>>\n’,p3)
p4 = np.delete(matrix, [0,1], 1) # 第1维度(列)第0、1行被删除
print(’>>>>p4>>>>\n’,p4)
结果：
p1>>>>
[[ 1 2 3 4]
[ 9 10 11 12]]
p2>>>>
[[ 1 3 4]
[ 5 7 8]
[ 9 11 12]]
p3>>>>
[ 1 3 4 5 6 7 8 9 10 11 12]
p4>>>>
[[ 3 4]
[ 7 8]
[11 12]]

insert()函数
import numpy as np
matrix = [[1,2,3,4],[5,6,7,8],[9,10,11,12]]
q1 = np.insert(matrix, 1, [1,1,1,1], 0) # 第0维度(行)第1行添加[1,1,1,1]
print(’>>>>q1>>>>\n’,q1)
q2 = np.insert(matrix, 0, [1,1,1], 1) # 第1维度(列)第0列添加1,1,1
print(’>>>>q2>>>>\n’,q2)
q3 = np.insert(matrix, 3, [1,1,1,1], 0) # 第0维度(行)第3行添加[1,1,1,1]
print(’>>>>q3>>>>\n’,q3)
结果
q1>>>>
[[ 1 2 3 4]
[ 1 1 1 1]
[ 5 6 7 8]
[ 9 10 11 12]]
q2>>>>
[[ 1 1 2 3 4]
[ 1 5 6 7 8]
[ 1 9 10 11 12]]
q3>>>>
[[ 1 2 3 4]
[ 5 6 7 8]
[ 9 10 11 12]
[ 1 1 1 1]]

append()函数
import numpy as np
matrix = [[1,2,3,4],[5,6,7,8],[9,10,11,12]]
m1 = np.append(matrix,[[1,1,1,1]],axis=0)
print(’>>>>m1>>>>\n’,m1)
m2 = np.append(matrix,[[1],[1],[1]],axis=1)
print(’>>>>m2>>>>\n’,m2)
m3 = np.append(matrix,[1,1,1,1])
print(’>>>>m3>>>>\n’,m3)
结果
m1>>>>
[[ 1 2 3 4]
[ 5 6 7 8]
[ 9 10 11 12]
[ 1 1 1 1]]
m2>>>>
[[ 1 2 3 4 1]
[ 5 6 7 8 1]
[ 9 10 11 12 1]]
m3>>>>
[ 1 2 3 4 5 6 7 8 9 10 11 12 1 1 1 1]

np.random.choice(a, size, replace, p)
import numpy as np
a1 = np.random.choice(7,5) # 从0~7中随机选择5个数组成一维数组
a1
array([3, 1, 0, 1, 4])
a2 = np.random.choice([0,1,2,3,4,5,6],5) # 从给定list中随机选择5个数组成一维数组
a2
array([1, 6, 6, 6, 4])
a3 = np.random.choice(np.array([0,1,2,3,4,5,6]),5) # 将list换成array数组依然可以运行，效果一致
a3
array([3, 0, 5, 5, 3])
a4 = np.random.choice([0,1,2,3,4,5,6],5,replace=False) # 上述均有重复，将replace设置为False，即可按要求没有重复的选取
a4
array([0, 4, 3, 6, 5])
a5 = np.random.choice(np.array([0,1,2,3,4,5,6]),5,p=[0.1,0.1,0.1,0.1,0.1,0.1,0.4])
a5
array([6, 3, 2, 6, 6])

np.argmax(a, axis=None, out=None)
import numpy as np
a = np.array([[1,1,1],[2,2,2],[0,3,6]])
a
array([[1, 1, 1],
[2, 2, 2],
[0, 3, 6]])
b1 = np.argmax(a) # 将数组a拉平，最大值索引为9(初始索引为0)
b1
8
b2 = np.argmax(a, axis=0) # 按列选取最大值的索引
b2
array([1, 2, 2], dtype=int64)
b3 = np.argmax(a, axis=1) # 按行选取最大值的索引
b3
array([0, 0, 2], dtype=int64)

星号(*)的作用
import numpy as np
y1 = np.linspace(-10.0,10.0) # 默认生成50个数据
y1
array([-10. , -9.59183673, -9.18367347, -8.7755102 ,
-8.36734694, -7.95918367, -7.55102041, -7.14285714,
-6.73469388, -6.32653061, -5.91836735, -5.51020408,
-5.10204082, -4.69387755, -4.28571429, -3.87755102,
-3.46938776, -3.06122449, -2.65306122, -2.24489796,
-1.83673469, -1.42857143, -1.02040816, -0.6122449 ,
-0.20408163, 0.20408163, 0.6122449 , 1.02040816,
1.42857143, 1.83673469, 2.24489796, 2.65306122,
3.06122449, 3.46938776, 3.87755102, 4.28571429,
4.69387755, 5.10204082, 5.51020408, 5.91836735,
6.32653061, 6.73469388, 7.14285714, 7.55102041,
7.95918367, 8.36734694, 8.7755102 , 9.18367347,
9.59183673, 10. ])
y2 = np.linspace(1,10,10) # 生成10个数据，包括首尾
y2
array([ 1., 2., 3., 4., 5., 6., 7., 8., 9., 10.])
y3 = np.linspace(1,10,10,endpoint=False) # 不包括尾部数据
y3
array([1. , 1.9, 2.8, 3.7, 4.6, 5.5, 6.4, 7.3, 8.2, 9.1])
y4= np.linspace(1, 10, 6, retstep=True) # 将步长与结果的数组放入一个list、
y4
(array([ 1. , 2.8, 4.6, 6.4, 8.2, 10. ]), 1.8)

拉平操作 ravel()和faltten()及reshape(1,-1)的区别联系(补充[None,:]操作)
import numpy as np
x = np.array([[1,2,3],[4,5,6],[1,2,3]])
x.flatten()
array([1, 2, 3, 4, 5, 6, 1, 2, 3]) # 拉平
x.ravel()
array([1, 2, 3, 4, 5, 6, 1, 2, 3])
x.ravel(‘F’)
array([1, 4, 1, 2, 5, 2, 3, 6, 3]) # 按列拉平
x.flatten(‘F’)
array([1, 4, 1, 2, 5, 2, 3, 6, 3])
x.flatten()[1] = 20
x
array([[1, 2, 3],
[4, 5, 6],
[1, 2, 3]])
x.ravel()[1] = 20
x
array([[ 1, 20, 3],
[ 4, 5, 6],
[ 1, 2, 3]])
x.reshape(1,-1) # 注意结果仍然是二维
array([[1, 2, 3, 4, 5, 6, 1, 2, 3]]) # 这里有两个方括号
x = np.array([1,2,3,6,7,8]) # 注意操作的是数组，即原x是数组
***>>> x[None,:] # 转成行向量(二维矩阵)
array([[1, 2, 3, 6, 7, 8]])
x[:,None] # 转成列向量(二维矩阵)***
array([[1],
[2],[3],[6],[7],[8]])
x[np.newaxis, :] # np.newaxis与None用法一致
array([[1, 2, 3, 6, 7, 8]])

np.prod() 计算元素乘积
x = np.array([[1,2,3],[2,3,4]])
np.prod(x)
144
np.prod(x,axis=1)
array([ 6, 24])
np.prod(x,axis=0)
array([ 2, 6, 12])

把矩阵大于或小于N的元素置M的技巧
import numpy as np
x = np.array([[1,2,3],[-3,2,4],[5,-2,9]])
x
array([[ 1, 2, 3],
[-3, 2, 4],
[ 5, -2, 9]])
y1 = np.maximum(0,x) # 把小于0的元素置0，比改变x的值
y1
array([[1, 2, 3],
[0, 2, 4],
[5, 0, 9]])
y2 = np.minimum(0,x) # 把大于0的元素置0，不改变x的值
y2
array([[ 0, 0, 0],
[-3, 0, 0],
[ 0, -2, 0]])
x1 = x.copy()
x1
array([[ 1, 2, 3],
[-3, 2, 4],
[ 5, -2, 9]])
x1[x1 < 0] = 0 # 把小于0的元素置0，改变x1的值
x1
array([[1, 2, 3],
[0, 2, 4],
[5, 0, 9]])
x2 = x.copy()
x2[x2 > 0] = 0 # 把大于0的元素置0，改变x2的值
x2
array([[ 0, 0, 0],
[-3, 0, 0],
[ 0, -2, 0]])

numpy中的矩阵copy问题
import numpy as np
x = np.array([[1,2,3],[-3,2,4],[5,-2,9]])
x
array([[ 1, 2, 3],
[-3, 2, 4],
[ 5, -2, 9]])
x1 = x.copy() # copy()，开辟新地址
x1[x1 > 0] = 0
x1
array([[ 0, 0, 0],
[-3, 0, 0],
[ 0, -2, 0]])
x # x不变
array([[ 1, 2, 3],
[-3, 2, 4],
[ 5, -2, 9]])
x2 = x # 直接等于，未开辟新地址，x2与x相关联
x2
array([[ 1, 2, 3],
[-3, 2, 4],
[ 5, -2, 9]])
x2[x2>0] = 0
x2
array([[ 0, 0, 0],
[-3, 0, 0],
[ 0, -2, 0]])
x # x也改变
array([[ 0, 0, 0],
[-3, 0, 0],
[ 0, -2, 0]])
x = np.array([[1,2,3],[-3,2,4],[5,-2,9]])
x3 = x[2] # 取x的第3行
x3
array([ 5, -2, 9])
x3[2] = 100 # 将x3第3个元素置100
x # x中对应的元素置也被置成100了
array([[ 1, 2, 3],
[ -3, 2, 4],
[ 5, -2, 100]])

np.zeros_like()构造全零矩阵，无需指定大小
import numpy as np
x = np.array([[1,2,3],[4,5,6]])
np.zeros_like(x) # 生成一个和x大小相同的全零矩阵
array([[0, 0, 0],
[0, 0, 0]])
1
2
3
4
5

产生随机浮点数
import numpy as np
n = np.random.rand(3,4)
n
array([[0.11502462, 0.3503468 , 0.6206656 , 0.35172063],
[0.66367565, 0.10195004, 0.22708003, 0.9318615 ],
[0.77946053, 0.81804715, 0.2272248 , 0.17736476]])

np.random.randn() 生成正太分布矩阵
import numpy as np
x = np.random.randn(2,3)
x
array([[-0.09382672, -0.97939614, 0.0222352 ],
[ 0.15060188, -0.01107968, 1.08374427]])
y = np.multiply(0.1,np.random.randn(2,3))+0.5 # 一般正太分布
y
array([[0.49305173, 0.36802044, 0.48699281],
[0.45197275, 0.53837051, 0.60022348]])

np.randm.randint() 生成离散均匀分布的整数值组成的矩阵
import numpy as np
z = np.random.randint(2,9,(2,3))
z
array([[2, 3, 6],
[8, 2, 8]])
m = np.random.randint(9,size = (2,3))
m
array([[4, 0, 2],
[5, 2, 8]])

python 断言 assert
x = ‘You are right’
type(x)
<class ‘str’>
assert type(x)==str, ‘x is not str’
x = [1,2,3]
type(x)
<class ‘list’>
assert type(x)==str, ‘x is not str’
Traceback (most recent call last):
File “”, line 1, in
AssertionError: x is not str

np.pad()函数进行padding操作
import numpy as np
A = np.arange(95,99).reshape(2,2)
A
array([[95, 96], [97, 98]])
np.pad(A,((3,2),(2,3)),‘constant’,constant_values = (0,0))
array([[ 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 95, 96, 0, 0, 0],
[ 0, 0, 97, 98, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0],
[ 0, 0, 0, 0, 0, 0, 0]])
说明： 在矩阵行的上部分添加3行，下部分添加2行，在列的左部分添加2列，右部分添加3列。
b = np.array([[[1,2],[3,4]],[[3,4],[7,8]],[[4,5],[1,2]]])
b
array([[[1, 2],
[3, 4]],
   [[3, 4],
    [7, 8]],
   [[4, 5],
    [1, 2]]])
1
2
3
4
5
np.pad(b, ((0,0),(1,1),(1,1)), ‘constant’, constant_values = 0)
array([[[0, 0, 0, 0],
[0, 1, 2, 0],
[0, 3, 4, 0],
[0, 0, 0, 0]],
   [[0, 0, 0, 0],
    [0, 3, 4, 0],
    [0, 7, 8, 0],
    [0, 0, 0, 0]],
   [[0, 0, 0, 0],
    [0, 4, 5, 0],
    [0, 1, 2, 0],
    [0, 0, 0, 0]]])

numpy.empty() 创建指定形状和数据类型且未初始化的数组
import numpy as np
x = np.empty([3,2], dtype = int)
print (x)

输出：
[[ 6917529027641081856 5764616291768666155]
[ 6917529027641081859 -5764598754299804209]
[ 4497473538 844429428932120]]
np.full 用于形成元素全为某元素的矩阵
import numpy as np
c = np.array([[1,2],[3,4]])
c
array([[1, 2],
[3, 4]])
c.astype(np.float32)
array([[1., 2.],
[3., 4.]], dtype=float32)

np.meshgrid() 快速生成网格
import numpy as np
x = np.array([1,3,5])
y = np.array([4,6])
XX,YY = np.meshgrid(x,y)
XX
array([[1, 3, 5],
[1, 3, 5]])
YY
array([[4, 4, 4],
[6, 6, 6]])

np.hstack() 和 np.vstack() 用于堆叠矩阵
import numpy as np
x = np.array([[3,4,5],[1,3,4]])
y = np.array([[1,1,1],[2,2,2]])
np.hstack((x,y)) # 水平堆叠
array([[3, 4, 5, 1, 1, 1],
[1, 3, 4, 2, 2, 2]])
np.vstack((x,y)) # 竖直堆叠
array([[3, 4, 5],
[1, 3, 4],
[1, 1, 1],
[2, 2, 2]])

numpy 取整
import numpy as np
a = np.array([0.125,0.568,5.688])
np.round(a) # 四舍五入取整， np.around 和 round 用法一致
array([0., 1., 6.])
np.round(a,decimals = 2) # 四舍五入保留2位小数
array([0.12, 0.57, 5.69])
np.floor(a) # 向下取整
array([0., 0., 5.])
np.ceil(a) # 向上取整
array([1., 1., 6.])

np.newaxis 在特定位置增加一个维度
import numpy as np
c = np.array([1,2,5,4])
c[:,np.newaxis]
array([[1],[2],[5],[4]])
c[np.newaxis,:]
array([[1, 2, 5, 4]])

python 广播机制
import numpy as np
a = np.array([[1,2,3],[4,5,6]])
a = np.array([[1,2,3,6],[4,5,6,6]])
a1 = a.reshape((1,2,4))
a1
array([[[1, 2, 3, 6],
[4, 5, 6, 6]]])
b = np.array([[3,4,5,6],[1,2,3,4],[4,5,5,5]])
b
array([[3, 4, 5, 6],
[1, 2, 3, 4],
[4, 5, 5, 5]])
b1 = b.reshape((1,3,4)).transpose((1,0,2))
b1
array([[[3, 4, 5, 6]],
   [[1, 2, 3, 4]],
   [[4, 5, 5, 5]]])
1
2
3
a1
array([[[1, 2, 3, 6],
[4, 5, 6, 6]]])
a1+b1
array([[[ 4, 6, 8, 12],
[ 7, 9, 11, 12]],
   [[ 2,  4,  6, 10],
    [ 5,  7,  9, 10]],
   [[ 5,  7,  8, 11],
    [ 8, 10, 11, 11]]])
将 a1 扩展成 [3,2,4] 成
[[[1, 2, 3, 6],
[4, 5, 6, 6]],
[[1, 2, 3, 6],
[4, 5, 6, 6]],
[[1, 2, 3, 6],
[4, 5, 6, 6]]]
将 b1 扩展成 [3,2,4] 成
[[[3, 4, 5, 6],
[3, 4, 5, 6]],
[[1, 2, 3, 4],
[1, 2, 3, 4]],
[[4, 5, 5, 5],
[4, 5, 5, 5]]]
扩展后的 a1 和 b1 相加得到
[[ 4, 6, 8, 12],
[ 7, 9, 11, 12]],
[[ 2, 4, 6, 10],
[ 5, 7, 9, 10]],
[[ 5, 7, 8, 11],
[ 8, 10, 11, 11]]]

numpy.transpose()转置
c = np.array([[[1,2,5],[3,4,6]],[[4,5,6],[7,8,9]]])
c
array([[[1, 2, 5],
[3, 4, 6]],
[[4, 5, 6],
[7, 8, 9]]])
c.transpose(1,0,2) # 将c的维度按照 第1维度，第0维度，第2维度的排序排成 第0,1,2维度
array([[[1, 2, 5],
[4, 5, 6]],
   [[3, 4, 6],
    [7, 8, 9]]])
1
2
如原来的 4 （4,5,6的4）本是 (1,0,0), 转换后变为 (0,1,0)
c.transpose(1,2,0) # 将c的维度按照 第1维度，第2维度，第0维度的排序排成 第0,1,2维度
array([[[1, 4],
[2, 5],
[5, 6]],[[3, 7],
[4, 8],[6, 9]]])

numpy 中双冒号的用法
import numpy as np
a = np.array([2,2,3,4,5,5,6,7])
a[0:7:2]
array([2, 3, 5, 6])
stop 值是默认尾部
import numpy as np
a = np.array([2,2,3,4,5,5,6,7])
a[0::2]
array([2, 3, 5, 6])
[::-1]
来将序列按倒序排列
a[::-1]
array([7, 6, 5, 5, 4, 3, 2, 2])
python 的内置函数slice(start, stop, step)
import numpy as np
a = np.array([2,2,3,4,5,5,6,7])
s = slice(0,7,2)
a[s]
array([2, 3, 5, 6])