python数据分析numpy基础之min求数组最小值

1 python数据分析numpy基础之min求数组最小值

python的numpy库的min()函数，用于计算沿指定轴(一个轴或多个轴)的最小值。

用法

min(a, axis=None, out=None, keepdims=<no value>, initial=<no value>, where=<no value>)

1.1 入参a

numpy.min()的入参a，为必选入参，可以为数组、列表、元组。

>>> import numpy as np
# 入参a为列表
>>> np.min([1,2,3,-1,-5])
-5
# 入参a为元组
>>> np.min((1,2,3,-1,-5))
-5
# 入参a为数组
>>> np.min(np.array((1,2,3,-1,-5)))
-5

1.2 入参axis为整数

numpy.min()的入参axis为可选入参，默认为None，表示求全部元素的最小值。

若axis=n为整数，则对指定轴n的元素求最小值。

若axis=负数，则-1对应最后一个轴，-2倒数第2个轴，依此类推。

>>> import numpy as np
>>> ar3=np.array([[[12, 13, 14, 15 ],
        [ 19,  25,  16,  7],
        [ 18,  29, 10, 31]],

       [[9,  11,  21,  3],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
>>> ar3
array([[[12, 13, 14, 15],
        [19, 25, 16,  7],
        [18, 29, 10, 31]],

       [[ 9, 11, 21,  3],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
# axis默认为None，整个数组的最小值
>>> np.min(ar3)
3
# axis=0，沿0轴的2个二维数组的每个位置都取最小值
>>> np.min(ar3,axis=0)
array([[ 9, 11, 14,  3],
       [16, 17, 16,  7],
       [18, 21, 10, 23]])
# axis=1，沿1轴的一维数组的每个位置都取最小值 
>>> np.min(ar3,axis=1)
array([[12, 13, 10,  7],
       [ 9, 11, 18,  3]])
# axis=2，沿2轴的每个位置都取最小值 
>>> np.min(ar3,axis=2)
array([[12,  7, 10],
       [ 3, 16, 20]])
# 若axis=负数，则-1对应最后一个轴，-2倒数第2个轴，依此类推。
>>> np.min(ar3,axis=-1)
array([[12,  7, 10],
       [ 3, 16, 20]])

1.3 入参axis为元组

numpy.min()的入参axis若为轴的元组，则对多个轴求最小值。

axis=(m,n) 先沿m轴取最小值，再沿n轴取最小值

axis=(m,n)等效于axis=(n,m)。

>>> import numpy as np
>>> ar3=np.array([[[12, 13, 14, 15 ],
        [ 19,  25,  16,  7],
        [ 18,  29, 10, 31]],

       [[9,  11,  21,  3],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
>>> ar3
array([[[12, 13, 14, 15],
        [19, 25, 16,  7],
        [18, 29, 10, 31]],

       [[ 9, 11, 21,  3],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
# axis=(m,n)先沿m轴取最小值，再沿n轴取最小值
>>> np.min(ar3,axis=(0,1))
array([ 9, 11, 10,  3])
# axis=(m,n)等于# axis=(n,m)
>>> np.min(ar3,axis=(1,0))
array([ 9, 11, 10,  3])
>>> np.min(ar3,axis=(0,2))
array([ 3,  7, 10])
>>> np.min(ar3,axis=(1,2))
array([7, 3])

1.4 入参initial

numpy.min()的入参initial为可选入参，表示用initial替换调输出元素的中大于initial的值。

>>> import numpy as np
>>> ar3=np.array([[[12, 13, 14, 15 ],
        [ 19,  25,  16,  7],
        [ 18,  29, 10, 31]],

       [[9,  11,  21,  3],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
>>> ar3
array([[[12, 13, 14, 15],
        [19, 25, 16,  7],
        [18, 29, 10, 31]],

       [[ 9, 11, 21,  3],
        [16, 17, 18, 19],
        [20, 21, 22, 23]]])
# 输出大于initial，则替换为initial，否则不变
>>> np.min(ar3,axis=(0,1))
array([ 9, 11, 10,  3])
>>> np.min(ar3,axis=(0,1),initial=20)
array([ 9, 11, 10,  3])
>>> np.min(ar3,axis=(0,1),initial=2)
array([2, 2, 2, 2])
>>> np.min(ar3,axis=(0,1),initial=5)
array([5, 5, 5, 3])