numpy get started¶
导入numpy库，并查看numpy版本
In [1]:
import numpy as np
In [2]:
np.__version__
Out[2]:
'1.13.0'
In [3]:
import matplotlib.pyplot as plt
In [4]:
cat = plt.imread('cat.jpg')
In [5]:
print(cat)
[[[231 186 131] [232 187 132] [233 188 133] ..., [100 54 54] [ 92 48 47] [ 85 43 44]] [[232 187 132] [232 187 132] [233 188 133] ..., [100 54 54] [ 92 48 47] [ 84 42 43]] [[232 187 132] [233 188 133] [233 188 133] ..., [ 99 53 53] [ 91 47 46] [ 83 41 42]] ..., [[199 119 82] [199 119 82] [200 120 83] ..., [189 99 65] [187 97 63] [187 97 63]] [[199 119 82] [199 119 82] [199 119 82] ..., [188 98 64] [186 96 62] [188 95 62]] [[199 119 82] [199 119 82] [199 119 82] ..., [188 98 64] [188 95 62] [188 95 62]]]
In [6]:
type(cat)
Out[6]:
numpy.ndarray
In [7]:
cat.shape
Out[7]:
(456, 730, 3)
In [8]:
plt.imshow(cat)
plt.show()
In [ ]:
#请问电影是什么，nd.array 四维
#(x,456,760,3)
一、创建ndarray¶
1. 使用np.array()由python list创建¶
参数为列表： [1, 4, 2, 5, 3]
注意：
- numpy默认ndarray的所有元素的类型是相同的
- 如果传进来的列表中包含不同的类型，则统一为同一类型，优先级：str>float>int
In [9]:
l = [3,1,4,5,9,6]
n = np.array(l)
In [13]:
display(n,l)
array([3, 1, 4, 5, 9, 6])
[3, 1, 4, 5, 9, 6]
In [15]:
display(n.shape,l.shape)
-------------------------------------------------------------- AttributeError Traceback (most recent call last) <ipython-input-15-5eeacc6c47ae> in <module>() ----> 1 display(n.shape,l.shape) AttributeError: 'list' object has no attribute 'shape'
In [16]:
n2 = np.array([[3,4,7,1],[3,0,1,8],[2,4,6,8]])
display(n2.shape)
(3, 4)
In [17]:
n3 = np.array(['0',9.18,20])
n3
Out[17]:
array(['0', '9.18', '20'],
dtype='<U4')
In [18]:
n4 = np.array([1,2,3.14])
n4
Out[18]:
array([ 1. , 2. , 3.14])
2. 使用np的routines函数创建¶
包含以下常见创建方法：
- np.ones(shape, dtype=None, order='C')
In [22]:
one = np.ones((40,80,3),dtype=int)
In [23]:
display(one.shape)
(40, 80, 3)
- np.zeros(shape, dtype=float, order='C')
In [24]:
np.zeros((100,200,5))
Out[24]:
array([[[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
...,
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.]],
[[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
...,
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.]],
[[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
...,
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.]],
...,
[[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
...,
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.]],
[[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
...,
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.]],
[[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
...,
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.],
[ 0., 0., 0., 0., 0.]]])
- np.full(shape, fill_value, dtype=None, order='C')
In [25]:
np.full((20,30),fill_value=8.88)
Out[25]:
array([[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88],
[ 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88, 8.88,
8.88, 8.88, 8.88]])
- np.eye(N, M=None, k=0, dtype=float)
对角线为1其他的位置为0
In [26]:
#矩阵，满秩矩阵
np.eye(5)
# x + y = 10
# x - y = 5
# 1 1
# 1 -1
# 1 1
# 0 -2
# 1 0
# 0 -2
# 1 0
# 0 1
# x + y = 10
# 2x + 2y = 20
# 1 1
# 2 2
# 1 1
# 0 0
Out[26]:
array([[ 1., 0., 0., 0., 0.],
[ 0., 1., 0., 0., 0.],
[ 0., 0., 1., 0., 0.],
[ 0., 0., 0., 1., 0.],
[ 0., 0., 0., 0., 1.]])
- np.linspace(start, stop, num=50, endpoint=True, retstep=False, dtype=None)
In [31]:
# lin = linear 线性
# 00000000
# 11111111 = 127
np.linspace(start = 0,stop = 150,num = 50,endpoint=False,retstep=True,dtype=np.int8)
Out[31]:
(array([ 0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30,
33, 36, 39, 42, 45, 48, 51, 54, 57, 60, 63,
66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96,
99, 102, 105, 108, 111, 114, 117, 120, 123, 126, -127,
-124, -121, -118, -115, -112, -109], dtype=int8), 3.0)
- np.arange([start, ]stop, [step, ]dtype=None)
In [32]:
np.arange(0,100,step = 3)
Out[32]:
array([ 0, 3, 6, 9, 12, 15, 18, 21, 24, 27, 30, 33, 36, 39, 42, 45, 48,
51, 54, 57, 60, 63, 66, 69, 72, 75, 78, 81, 84, 87, 90, 93, 96, 99])
In [33]:
np.arange(10)
Out[33]:
array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
- np.random.randint(low, high=None, size=None, dtype='l')
In [36]:
#ndarray
np.random.randint(-100,100,size = (4,5))
Out[36]:
array([[-87, -29, -85, 15, -80],
[-94, -20, 8, 84, -94],
[-89, 59, -71, 9, 77],
[ 82, -13, 92, -50, 67]])
- np.random.randn(d0, d1, ..., dn)
标准正太分布
In [38]:
#d0,d1……dn
#dimensional 维度
np.random.randn(10,10)
Out[38]:
array([[ -5.46824594e-01, 1.07179174e+00, 1.77507447e+00,
2.25209098e-01, 5.19083202e-01, -6.76655557e-01,
1.99384931e+00, 1.16875392e+00, -5.53496016e-01,
-9.95607941e-01],
[ -1.34229240e+00, 1.82715427e+00, 1.68100791e-01,
1.04885582e-01, 2.71417935e+00, -2.58693080e-01,
1.04068515e-01, 8.10473766e-01, 1.90595706e-01,
5.59027533e-01],
[ -1.31158145e+00, -2.65147714e+00, -1.06652645e-01,
3.66054578e-01, 1.00488575e+00, 6.28278594e-01,
-5.87102941e-01, -5.19958644e-01, -6.07998002e-01,
4.47194822e-01],
[ 6.33424165e-01, 1.24501227e+00, 6.77243149e-01,
1.13717064e+00, -1.11405630e+00, 4.92695535e-01,
-4.50442972e-01, 4.80813688e-01, -3.93747118e-01,
3.67064763e-01],
[ 7.10941655e-01, -7.38304618e-01, -4.78201185e-01,
1.12340783e+00, 2.34967908e+00, 1.97657956e-01,
1.86172500e-02, 2.59809012e-01, 1.76131029e+00,
-7.42756403e-01],
[ -8.66588876e-02, -1.07209838e+00, -3.64836658e-01,
-8.40909028e-01, -7.37350991e-01, -8.90749448e-01,
-1.41449767e+00, -1.81072400e+00, 1.81216340e+00,
4.79779577e-01],
[ -1.56467346e-01, -3.26438621e-01, 5.77616121e-02,
-5.71762728e-01, 6.63998462e-01, -3.85444509e-01,
3.66929808e-01, 6.49857210e-01, -2.75195142e-01,
-5.01095259e-01],
[ 6.01485754e-01, -2.21271029e-02, 2.52285405e-02,
1.80423496e-01, -5.63482448e-05, 1.42370303e-01,
-1.00389352e+00, -1.45804209e-01, 8.44376609e-01,
1.23404325e+00],
[ 1.51250923e+00, 1.41495165e+00, -6.30276103e-01,
-4.76913259e-01, -1.58701741e+00, -9.16346290e-01,
-1.44218761e-01, -4.84195208e-01, -1.73369711e+00,
1.13364493e-01],
[ 1.27400732e+00, -1.46409799e+00, -1.22425105e+00,
-1.03935711e-01, -1.22836257e+00, 2.20029526e-01,
-1.63008922e-01, 8.41643451e-01, -1.69272378e-01,
1.41444618e+00]])
9)np.random.normal(loc=0.0, scale=1.0, size=None)
In [41]:
np.random.normal(175,100,56)
Out[41]:
array([ 269.55201826, 187.33873198, 120.12980915, 35.94753307,
100.90507802, 188.10840055, 193.31965054, 36.05935621,
140.17431892, 262.08854751, 157.21405498, 266.08319372,
343.02054984, 192.9958422 , 443.18707824, 224.91608598,
-54.74691554, 172.59002564, 204.49702642, 372.54441072,
126.86764549, 186.91726897, -3.66107589, 190.47178624,
142.05410227, 210.80513529, 293.25789963, 50.89529749,
313.27820574, -17.55536856, 78.67924885, 136.45928163,
191.91273968, 207.10742181, 39.29293149, 305.20744152,
346.83050746, 183.26777519, 83.58301573, -47.49004075,
213.37605839, 96.41939774, 201.9269295 , 110.29843417,
342.29758371, 162.90875179, 139.13639372, 226.59402494,
278.0888479 , 262.75948451, 49.66327788, 288.96514573,
146.53039311, 323.20823318, 123.20716854, 197.65838639])
- np.random.random(size=None)
生成0到1的随机数，左闭右开
In [42]:
np.random.random(size = 10)
Out[42]:
array([ 0.38379228, 0.13716473, 0.37084239, 0.10899928, 0.45899816,
0.30014622, 0.2898527 , 0.87834826, 0.59921499, 0.9087303 ])
使用随机数成成一张图片¶
In [43]:
image = np.random.random(size = (456,730,3))
In [44]:
plt.imshow(image)
plt.show()
In [45]:
plt.imshow(cat)
plt.show()
In [46]:
display(cat.shape,image.shape)
(456, 730, 3)
(456, 730, 3)
二、ndarray的属性¶
4个必记参数： ndim：维度 shape：形状（各维度的长度） size：总长度
dtype：元素类型
In [47]:
cat.ndim
Out[47]:
3
In [48]:
image.shape
Out[48]:
(456, 730, 3)
In [49]:
image.size
Out[49]:
998640
In [50]:
image.dtype
Out[50]:
dtype('float64')
三、ndarray的基本操作¶
1. 索引¶
一维与列表完全一致 多维时同理
In [51]:
n1 = np.array([1,2,4,7,9])
n1[3]
Out[51]:
7
In [53]:
n[0,1,1]
Out[53]:
8
In [54]:
n = np.random.randint(0,100,size = (4,5,6,2,3))
n
Out[54]:
array([[[[[55, 85, 59],
[76, 37, 8]],
[[94, 34, 7],
[ 4, 26, 71]],
[[59, 55, 8],
[44, 73, 16]],
[[18, 82, 66],
[89, 61, 48]],
[[90, 18, 49],
[10, 49, 47]],
[[24, 61, 57],
[17, 14, 47]]],
[[[96, 14, 71],
[57, 62, 3]],
[[33, 93, 9],
[14, 72, 48]],
[[95, 41, 50],
[ 5, 17, 86]],
[[ 3, 2, 39],
[74, 89, 58]],
[[70, 82, 95],
[50, 74, 11]],
[[36, 58, 60],
[97, 40, 49]]],
[[[72, 71, 76],
[ 3, 11, 83]],
[[42, 22, 76],
[27, 14, 24]],
[[85, 21, 63],
[86, 54, 27]],
[[10, 5, 74],
[96, 75, 87]],
[[65, 20, 68],
[27, 28, 40]],
[[37, 23, 86],
[43, 76, 59]]],
[[[48, 14, 32],
[80, 48, 8]],
[[39, 20, 43],
[83, 22, 3]],
[[79, 53, 68],
[35, 64, 97]],
[[79, 36, 44],
[47, 11, 72]],
[[37, 59, 51],
[65, 40, 14]],
[[86, 76, 44],
[42, 71, 95]]],
[[[ 8, 98, 49],
[96, 72, 14]],
[[93, 21, 31],
[ 7, 14, 1]],
[[29, 67, 79],
[36, 9, 87]],
[[99, 97, 96],
[62, 52, 35]],
[[66, 30, 80],
[35, 79, 41]],
[[82, 39, 27],
[39, 41, 26]]]],
[[[[46, 77, 21],
[54, 68, 96]],
[[46, 37, 96],
[45, 69, 22]],
[[59, 99, 27],
[60, 88, 1]],
[[98, 47, 26],
[87, 55, 20]],
[[45, 57, 10],
[15, 95, 33]],
[[16, 10, 88],
[40, 47, 51]]],
[[[99, 40, 4],
[95, 36, 64]],
[[ 7, 91, 52],
[17, 98, 51]],
[[18, 0, 96],
[95, 77, 49]],
[[71, 97, 51],
[12, 5, 64]],
[[27, 99, 83],
[76, 69, 24]],
[[80, 56, 73],
[52, 95, 76]]],
[[[37, 76, 43],
[12, 97, 39]],
[[19, 97, 95],
[84, 7, 36]],
[[58, 79, 86],
[76, 6, 56]],
[[42, 28, 58],
[74, 35, 65]],
[[88, 87, 16],
[72, 96, 46]],
[[73, 61, 30],
[98, 49, 24]]],
[[[74, 92, 30],
[95, 41, 50]],
[[24, 81, 46],
[24, 23, 6]],
[[67, 92, 77],
[59, 18, 93]],
[[65, 68, 38],
[83, 50, 72]],
[[86, 95, 68],
[99, 29, 58]],
[[54, 35, 72],
[80, 4, 73]]],
[[[69, 2, 21],
[74, 0, 91]],
[[17, 62, 56],
[ 7, 49, 51]],
[[35, 23, 87],
[12, 60, 2]],
[[18, 94, 10],
[25, 64, 38]],
[[55, 20, 37],
[35, 1, 50]],
[[75, 0, 93],
[38, 86, 92]]]],
[[[[30, 88, 19],
[ 5, 73, 51]],
[[17, 9, 48],
[43, 1, 15]],
[[82, 23, 99],
[66, 30, 64]],
[[89, 94, 9],
[ 7, 72, 6]],
[[93, 72, 20],
[39, 67, 38]],
[[95, 2, 47],
[49, 10, 56]]],
[[[ 7, 8, 96],
[46, 62, 92]],
[[70, 77, 79],
[94, 43, 31]],
[[68, 95, 58],
[15, 78, 75]],
[[31, 73, 94],
[72, 66, 47]],
[[54, 94, 44],
[44, 0, 92]],
[[92, 15, 17],
[95, 64, 29]]],
[[[18, 88, 13],
[25, 1, 37]],
[[45, 87, 96],
[65, 91, 92]],
[[11, 1, 63],
[35, 78, 54]],
[[90, 24, 29],
[98, 70, 62]],
[[93, 12, 45],
[91, 64, 87]],
[[68, 42, 39],
[29, 12, 61]]],
[[[11, 91, 13],
[65, 13, 13]],
[[12, 90, 76],
[72, 7, 45]],
[[59, 93, 32],
[ 3, 65, 55]],
[[35, 51, 90],
[49, 11, 36]],
[[42, 95, 81],
[18, 37, 55]],
[[74, 60, 9],
[14, 93, 64]]],
[[[45, 4, 69],
[35, 29, 31]],
[[91, 59, 11],
[ 9, 15, 14]],
[[39, 83, 34],
[42, 31, 83]],
[[23, 46, 47],
[46, 35, 62]],
[[22, 93, 80],
[22, 41, 12]],
[[68, 29, 16],
[53, 0, 62]]]],
[[[[27, 28, 82],
[55, 95, 34]],
[[94, 45, 41],
[92, 3, 10]],
[[50, 29, 40],
[77, 80, 35]],
[[72, 91, 88],
[20, 15, 70]],
[[10, 6, 31],
[27, 63, 36]],
[[89, 79, 83],
[21, 26, 94]]],
[[[72, 55, 13],
[57, 96, 34]],
[[62, 46, 53],
[12, 39, 56]],
[[62, 15, 46],
[80, 68, 47]],
[[77, 58, 15],
[30, 40, 99]],
[[59, 65, 23],
[75, 89, 48]],
[[95, 53, 64],
[11, 81, 20]]],
[[[88, 92, 24],
[47, 84, 38]],
[[92, 86, 72],
[65, 54, 45]],
[[34, 58, 49],
[91, 69, 27]],
[[75, 32, 42],
[36, 40, 66]],
[[36, 11, 41],
[81, 3, 60]],
[[91, 0, 66],
[49, 73, 43]]],
[[[40, 46, 26],
[41, 11, 50]],
[[20, 96, 7],
[56, 51, 2]],
[[76, 94, 46],
[93, 41, 90]],
[[67, 73, 27],
[ 8, 42, 46]],
[[50, 10, 92],
[ 6, 89, 8]],
[[68, 9, 6],
[31, 97, 45]]],
[[[76, 21, 70],
[34, 63, 32]],
[[14, 17, 11],
[ 9, 46, 3]],
[[82, 46, 23],
[13, 20, 27]],
[[39, 15, 37],
[23, 31, 32]],
[[72, 67, 83],
[66, 23, 14]],
[[84, 42, 5],
[31, 55, 21]]]]])
根据索引修改数据
2. 切片¶
一维与列表完全一致 多维时同理
In [55]:
n = np.random.randint(0,100,size = 10)
n
Out[55]:
array([46, 1, 53, 64, 73, 88, 74, 60, 99, 92])
In [57]:
n[3:6]
Out[57]:
array([64, 73, 88])
In [58]:
n[-2:]
Out[58]:
array([99, 92])
将数据反转，例如[1,2,3]---->[3,2,1]¶
In [59]:
n
Out[59]:
array([46, 1, 53, 64, 73, 88, 74, 60, 99, 92])
In [60]:
n[::-1]
Out[60]:
array([92, 99, 60, 74, 88, 73, 64, 53, 1, 46])
In [63]:
n[::-2]
Out[63]:
array([92, 60, 88, 64, 1])
两个::进行切片
In [ ]:
3. 变形¶
使用reshape函数，注意参数是一个tuple！
In [65]:
cat.size
Out[65]:
998640
In [71]:
n = np.random.randint(0,255,size = 998640)
n
Out[71]:
array([123, 192, 72, ..., 128, 245, 181])
In [76]:
image2 = n.reshape((456,730,4))
-------------------------------------------------------------- ValueError Traceback (most recent call last) <ipython-input-76-9102b6480bfd> in <module>() ----> 1 image2 = n.reshape((456,730,4)) ValueError: cannot reshape array of size 998640 into shape (456,730,4)
In [72]:
image2
Out[72]:
array([[[204, 170, 115],
[235, 183, 19],
[111, 136, 113],
...,
[ 91, 166, 47],
[ 62, 196, 199],
[105, 203, 61]],
[[ 82, 199, 56],
[103, 31, 234],
[ 74, 228, 172],
...,
[133, 113, 54],
[ 21, 33, 66],
[138, 113, 154]],
[[ 16, 150, 252],
[241, 211, 97],
[ 62, 16, 113],
...,
[ 60, 198, 193],
[217, 152, 236],
[153, 226, 113]],
...,
[[242, 198, 35],
[ 27, 220, 30],
[ 96, 194, 192],
...,
[ 79, 91, 224],
[181, 117, 2],
[143, 9, 211]],
[[ 62, 129, 0],
[ 7, 205, 61],
[140, 51, 9],
...,
[243, 14, 92],
[ 63, 100, 101],
[118, 81, 124]],
[[ 17, 88, 179],
[145, 96, 166],
[ 15, 154, 33],
...,
[ 59, 173, 234],
[184, 150, 243],
[106, 247, 165]]])
In [74]:
image3 = image2.astype(float)
In [75]:
plt.imshow(image3)
plt.show()
4. 级联¶
- np.concatenate()
级联需要注意的点：
- 级联的参数是列表：一定要加中括号或小括号
- 维度必须相同
- 形状相符
- 【重点】级联的方向默认是shape这个tuple的第一个值所代表的维度方向
- 可通过axis参数改变级联的方向
In [83]:
n1 = np.random.randint(0,100,size = (4,2))
n2 = np.random.randint(-100,0,size = (4,6))
In [84]:
display(n1,n2)
array([[43, 63],
[48, 58],
[57, 58],
[29, 63]])
array([[-14, -35, -38, -31, -35, -47],
[ -1, -56, -55, -40, -91, -2],
[-52, -81, -40, -35, -85, -5],
[-10, -40, -16, -33, -46, -21]])
In [86]:
np.concatenate((n1,n2),axis = 1)
Out[86]:
array([[ 43, 63, -14, -35, -38, -31, -35, -47],
[ 48, 58, -1, -56, -55, -40, -91, -2],
[ 57, 58, -52, -81, -40, -35, -85, -5],
[ 29, 63, -10, -40, -16, -33, -46, -21]])
In [80]:
n1.ndim
Out[80]:
2
In [81]:
n1.shape
Out[81]:
(2, 3)
In [82]:
np.concatenate((n1,n2),axis = 1)
-------------------------------------------------------------- ValueError Traceback (most recent call last) <ipython-input-82-8b08dc61df7f> in <module>() ----> 1 np.concatenate((n1,n2),axis = 1) ValueError: all the input array dimensions except for the concatenation axis must match exactly
In [87]:
#cat image
image2 = np.concatenate((cat,image),axis = 0)
image2.shape
Out[87]:
(912, 730, 3)
In [89]:
cat
Out[89]:
array([[[231, 186, 131],
[232, 187, 132],
[233, 188, 133],
...,
[100, 54, 54],
[ 92, 48, 47],
[ 85, 43, 44]],
[[232, 187, 132],
[232, 187, 132],
[233, 188, 133],
...,
[100, 54, 54],
[ 92, 48, 47],
[ 84, 42, 43]],
[[232, 187, 132],
[233, 188, 133],
[233, 188, 133],
...,
[ 99, 53, 53],
[ 91, 47, 46],
[ 83, 41, 42]],
...,
[[199, 119, 82],
[199, 119, 82],
[200, 120, 83],
...,
[189, 99, 65],
[187, 97, 63],
[187, 97, 63]],
[[199, 119, 82],
[199, 119, 82],
[199, 119, 82],
...,
[188, 98, 64],
[186, 96, 62],
[188, 95, 62]],
[[199, 119, 82],
[199, 119, 82],
[199, 119, 82],
...,
[188, 98, 64],
[188, 95, 62],
[188, 95, 62]]], dtype=uint8)
In [91]:
image3
Out[91]:
array([[[ 204., 170., 115.],
[ 235., 183., 19.],
[ 111., 136., 113.],
...,
[ 91., 166., 47.],
[ 62., 196., 199.],
[ 105., 203., 61.]],
[[ 82., 199., 56.],
[ 103., 31., 234.],
[ 74., 228., 172.],
...,
[ 133., 113., 54.],
[ 21., 33., 66.],
[ 138., 113., 154.]],
[[ 16., 150., 252.],
[ 241., 211., 97.],
[ 62., 16., 113.],
...,
[ 60., 198., 193.],
[ 217., 152., 236.],
[ 153., 226., 113.]],
...,
[[ 242., 198., 35.],
[ 27., 220., 30.],
[ 96., 194., 192.],
...,
[ 79., 91., 224.],
[ 181., 117., 2.],
[ 143., 9., 211.]],
[[ 62., 129., 0.],
[ 7., 205., 61.],
[ 140., 51., 9.],
...,
[ 243., 14., 92.],
[ 63., 100., 101.],
[ 118., 81., 124.]],
[[ 17., 88., 179.],
[ 145., 96., 166.],
[ 15., 154., 33.],
...,
[ 59., 173., 234.],
[ 184., 150., 243.],
[ 106., 247., 165.]]])
In [92]:
image4 = np.concatenate((cat,image3),axis = 0)
image4.shape
Out[92]:
(912, 730, 3)
In [106]:
cat[:2]
Out[106]:
array([[[231, 186, 131],
[232, 187, 132],
[233, 188, 133],
...,
[100, 54, 54],
[ 92, 48, 47],
[ 85, 43, 44]],
[[232, 187, 132],
[232, 187, 132],
[233, 188, 133],
...,
[100, 54, 54],
[ 92, 48, 47],
[ 84, 42, 43]]], dtype=uint8)
In [107]:
image4[:2]
Out[107]:
array([[[ 231., 186., 131.],
[ 232., 187., 132.],
[ 233., 188., 133.],
...,
[ 100., 54., 54.],
[ 92., 48., 47.],
[ 85., 43., 44.]],
[[ 232., 187., 132.],
[ 232., 187., 132.],
[ 233., 188., 133.],
...,
[ 100., 54., 54.],
[ 92., 48., 47.],
[ 84., 42., 43.]]])
In [109]:
cat.dtype
Out[109]:
dtype('uint8')
In [112]:
plt.imshow(image4.astype('uint8'))
plt.show()
- np.hstack与np.vstack
水平级联与垂直级联,处理自己，进行维度的变更
In [114]:
n = np.random.randint(0,100,size = (4,5))
n
Out[114]:
array([[28, 15, 65, 42, 20],
[ 7, 22, 98, 86, 58],
[14, 49, 31, 81, 80],
[97, 67, 22, 73, 2]])
In [115]:
#h = horizontal 水平
np.hstack(n)
Out[115]:
array([28, 15, 65, 42, 20, 7, 22, 98, 86, 58, 14, 49, 31, 81, 80, 97, 67,
22, 73, 2])
In [116]:
n = np.random.randint(0,10,size = 5)
n
Out[116]:
array([8, 2, 7, 3, 3])
In [117]:
np.vstack(n)
Out[117]:
array([[8],
[2],
[7],
[3],
[3]])
5. 切分¶
与级联类似，三个函数完成切分工作：
- np.split
- np.vsplit
- np.hsplit
In [127]:
n = np.random.randint(0,100,size = (4,6))
n
Out[127]:
array([[33, 46, 53, 7, 56, 43],
[98, 40, 12, 55, 18, 64],
[47, 53, 69, 50, 69, 7],
[34, 91, 32, 52, 83, 38]])
In [130]:
np.vsplit(n,(1,2))
Out[130]:
[array([[33, 46, 53, 7, 56, 43]]),
array([[98, 40, 12, 55, 18, 64]]),
array([[47, 53, 69, 50, 69, 7],
[34, 91, 32, 52, 83, 38]])]
In [131]:
np.hsplit(n,(2,4))
Out[131]:
[array([[33, 46],
[98, 40],
[47, 53],
[34, 91]]), array([[53, 7],
[12, 55],
[69, 50],
[32, 52]]), array([[56, 43],
[18, 64],
[69, 7],
[83, 38]])]
In [118]:
cat.shape
Out[118]:
(456, 730, 3)
In [ ]:
456
730
In [124]:
s_result = np.split(cat,2,axis = 1)
In [125]:
len(s_result)
Out[125]:
2
In [126]:
plt.imshow(s_result[0])
plt.show()
6. 副本¶
所有赋值运算不会为ndarray的任何元素创建副本。对赋值后的对象的操作也对原来的对象生效。
In [132]:
n = np.array([1,2,3])
In [135]:
n[2] = 10
In [136]:
n
Out[136]:
array([ 1, 2, 10])
可使用copy()函数创建副本
In [137]:
n2 = n.copy()
In [139]:
n2[0] = 1024
n2
Out[139]:
array([1024, 2, 10])
In [140]:
n
Out[140]:
array([ 1, 2, 10])
四、ndarray的聚合操作¶
1. 求和np.sum¶
2. 最大最小值：np.max/ np.min¶
同理
3. 其他聚合操作¶
Function Name NaN-safe Version Description
np.sum np.nansum Compute sum of elements
np.prod np.nanprod Compute product of elements
np.mean np.nanmean Compute mean of elements
np.std np.nanstd Compute standard deviation
np.var np.nanvar Compute variance
np.min np.nanmin Find minimum value
np.max np.nanmax Find maximum value
np.argmin np.nanargmin Find index of minimum value
np.argmax np.nanargmax Find index of maximum value
np.median np.nanmedian Compute median of elements
np.percentile np.nanpercentile Compute rank-based statistics of elements
np.any N/A Evaluate whether any elements are true
np.all N/A Evaluate whether all elements are true
np.power 幂运算
In [144]:
cat.shape
Out[144]:
(456, 730, 3)
In [152]:
cat2 = cat.reshape((-1,3))
cat2.shape
Out[152]:
(332880, 3)
In [153]:
cat3 = cat.reshape((456*730,3))
cat3.shape
Out[153]:
(332880, 3)
In [154]:
cat3.max(axis = 0)
Out[154]:
array([255, 242, 219], dtype=uint8)
In [149]:
max_cat = cat.max(axis = (0,1))
In [150]:
max_cat
Out[150]:
array([255, 242, 219], dtype=uint8)
In [151]:
max_cat.shape
Out[151]:
(3,)
In [142]:
cat.min()
Out[142]:
0
In [ ]:
In [ ]:
np.sum 和 np.nansum 的区别 nan not a number
In [ ]:
操作文件¶
使用pandas打开文件president_heights.csv 获取文件中的数据
In [ ]:
五、ndarray的矩阵操作¶
1. 基本矩阵操作¶
- 算术运算符：
- 加减乘除
- 矩阵积np.dot()
In [ ]:
2. 广播机制¶
【重要】ndarray广播机制的两条规则
- 规则一：为缺失的维度补1
- 规则二：假定缺失元素用已有值填充
例1： m = np.ones((2, 3)) a = np.arange(3) 求M+a
In [ ]:
例2： a = np.arange(3).reshape((3, 1)) b = np.arange(3) 求a+b
习题 a = np.ones((4, 1)) b = np.arange(4) 求a+b
六、ndarray的排序¶
小测验： 使用以上所学numpy的知识，对一个ndarray对象进行选择排序。
def Sortn(x):
代码越短越好
In [ ]:
In [ ]:
2. 部分排序¶
np.partition(a,k)
有的时候我们不是对全部数据感兴趣，我们可能只对最小或最大的一部分感兴趣。
- 当k为正时，我们想要得到最小的k个数
- 当k为负时，我们想要得到最大的k个数