____tz_zs

之前把 numpy 資料寫在了同一篇部落格裡，發現非常難以查閱，於是按功能切分開來。

運算

• ndarray.min() / np.min(ndarray)
• ndarray.max() / np.max(ndarray)
• ndarray.mean() / np.mean(ndarray)
• ndarray.sum() / np.sum(ndarray)
• np.exp()
• np.sqrt() 開根號
• np.sort() 排序
• np.argsort() 排序的索引

.

# -*- coding: utf-8 -*-
"""
@author: tz_zs

ndarray.min()
ndarray.max()
ndarray.mean()
ndarray.sum()
ndarray.argmax() 最大值的索引
np.exp()
np.sqrt() 開根號
np.floor() 地板值
np.sort() 排序
np.argsort() 排序的索引
"""

import numpy as np

n = np.array([[5, 10, 15],
              [20, 25, 30],
              [35, 40, 45]])
print(n.min())  # 5
print(n.max())  # 5
print(n.mean())  # 25.0
print(n.sum())  # 225
# 指定所操作的維度，axis=0 按列，axis=1 按行
print(n.sum(axis=0))  # [60 75 90]
print(n.sum(axis=1))  # [ 30  75 120]

print('-' * 20)
n1 = np.arange(3)
print(n1)
print(np.exp(n1))
print(np.sqrt(n1))
'''
[0 1 2]
[ 1.          2.71828183  7.3890561 ]
[ 0.          1.          1.41421356]
'''

# np.floor()
print('-' * 20)
n2 = np.random.random((2, 3)) * 10
n3 = np.floor(n2)
print(n2)
print(n3)
'''
[[ 6.46353334  8.29433697  4.78221334]
 [ 2.95695022  3.80006904  3.00482368]]
[[ 6.  8.  4.]
 [ 2.  3.  3.]]
'''

# np.sort() 排序
print('-' * 20)
a = np.array([[4, 3, 5], [1, 6, 1]])
print(a)
'''
[[4 3 5]
 [1 6 1]]
'''
b = np.sort(a, axis=1)
print(b)
'''
[[3 4 5]
 [1 1 6]]
'''
a.sort(axis=1)
print(a)
'''
[[3 4 5]
 [1 1 6]]
'''

# np.argsort() 排序的索引
print('-' * 20)
a2 = np.array([4, 3, 1, 2])
j = np.argsort(a2)
print(j)  # [2 3 1 0]
print(a2[j])  # [1 2 3 4]
 

·

矩陣乘法

ndarray1 * ndarray2
ndarray1.dot(ndarray2)

np.dot(ndarray1, ndarray2)

# -*- coding: utf-8 -*-
"""
@author: tz_zs
矩陣乘法
ndarray1 * ndarray2
ndarray1.dot(ndarray2)
np.dot(ndarray1, ndarray2)
"""

import numpy as np

# The matrix product can be performed using the dot function or method
n1 = np.array([[1, 1],
               [0, 1]])
n2 = np.array([[2, 0],
               [3, 4]])
print(n1)
print(n2)
'''
[[1 1]
 [0 1]]
[[2 0]
 [3 4]]
'''

print(n1 * n2)
'''
[[2 0]
 [0 4]]
'''

print(n1.dot(n2))
print(np.dot(n1, n2))
'''
[[5 4]
 [3 4]]
[[5 4]
 [3 4]]
'''
 

·

dot

numpy.dot(a, b, out=None)

a : 第一個引數，b : 第二個引數，out : 控制輸出的引數 

此函式對於一維矩陣，計算的是點積（向量乘法），對於二維矩陣，計算矩陣乘積。

.

拼接/切分 ndarray

拼接 ndarray 物件

• np.hstack()  # 橫著拼接
• np.vstack()  # 豎著拼接

切分 ndarray 物件

• np.vsplit() # 豎著切

# -*- coding: utf-8 -*-
"""
@author: tz_zs

拼接 ndarray 物件
np.hstack()  # 橫著拼接
np.vstack()  # 豎著拼接

切分 ndarray 物件
np.vsplit() 豎著切
"""

import numpy as np

n1 = np.array([[10, 11],
               [12, 13]])
n2 = np.array([[20, 21],
               [22, 23]])

hstack = np.hstack((n1, n2))  # 橫著拼接
print(hstack)
'''
[[10 11 20 21]
 [12 13 22 23]]
'''

vstack = np.vstack((n1, n2))  # 豎著拼接
print(vstack)
'''
[[10 11]
 [12 13]
 [20 21]
 [22 23]]
'''

# np.hsplit() 橫著切
n3 = np.floor(10 * np.random.random((2, 12)))
print(n3)
'''
[[ 6.  1.  1.  9.  9.  5.  2.  3.  4.  4.  0.  2.]
 [ 6.  3.  7.  7.  7.  9.  0.  2.  3.  0.  4.  8.]]
'''
print(np.hsplit(n3, 3))
'''
[array([[ 6.,  1.,  1.,  9.],
       [ 6.,  3.,  7.,  7.]]), 
 array([[ 9.,  5.,  2.,  3.],
       [ 7.,  9.,  0.,  2.]]), 
 array([[ 4.,  4.,  0.,  2.],
       [ 3.,  0.,  4.,  8.]])]
'''
print(np.hsplit(n3, (3, 5)))  # 指定切分位置
'''
[array([[ 6.,  1.,  1.],
       [ 6.,  3.,  7.]]), 
 array([[ 9.,  9.],
       [ 7.,  7.]]), 
 array([[ 5.,  2.,  3.,  4.,  4.,  0.,  2.],
       [ 9.,  0.,  2.,  3.,  0.,  4.,  8.]])]
'''

# np.vsplit() 豎著切
n4 = n3.T
print(n4)
'''
[[ 6.  6.]
 [ 1.  3.]
 [ 1.  7.]
 [ 9.  7.]
 [ 9.  7.]
 [ 5.  9.]
 [ 2.  0.]
 [ 3.  2.]
 [ 4.  3.]
 [ 4.  0.]
 [ 0.  4.]
 [ 2.  8.]]
'''
print(np.vsplit(n4, 3))
'''
[array([[ 6.,  6.],
       [ 1.,  3.],
       [ 1.,  7.],
       [ 9.,  7.]]), 
array([[ 9.,  7.],
       [ 5.,  9.],
       [ 2.,  0.],
       [ 3.,  2.]]), 
array([[ 4.,  3.],
       [ 4.,  0.],
       [ 0.,  4.],
       [ 2.,  8.]])]
'''
 

·