随机数函数

• np.random.randint()

>>> b= np.random.randint(100,200,(3,4))		# 生成100-200之间随机数的3行4列数组
>>> b
array([[134, 183, 138, 181],
       [132, 107, 194, 159],
       [154, 170, 130, 177]])

• np.random.seed(s)

>>> np.random.randint(10,20,(3,4))			# 随机数种子，s为给定种子值
array([[13, 16, 16, 10],
       [19, 18, 14, 17],
       [10, 10, 17, 11]])
>>> np.random.seed(5)
>>> np.random.randint(10,20,(3,4))
array([[13, 16, 16, 10],
       [19, 18, 14, 17],
       [10, 10, 17, 11]])

使用共同的seed值，则随机数数值相同
设置的seed值只能有效一次

• random.rand()

>>> np.random.rand(3,4)						# 生成3行4列的数组，随机数为浮点数 范围[0,1)，均匀分布
array([[0.08074127, 0.7384403 , 0.44130922, 0.15830987],
       [0.87993703, 0.27408646, 0.41423502, 0.29607993],
       [0.62878791, 0.57983781, 0.5999292 , 0.26581912]])
>>> np.random.rand(3,3,4)						# 生成3维3行4列的数组，随机数为浮点数 范围[0,1)，均匀分布
array([[[0.28468588, 0.25358821, 0.32756395, 0.1441643 ],
        [0.16561286, 0.96393053, 0.96022672, 0.18841466],
        [0.02430656, 0.20455555, 0.69984361, 0.77951459]],

       [[0.02293309, 0.57766286, 0.00164217, 0.51547261],
        [0.63979518, 0.9856244 , 0.2590976 , 0.80249689],
        [0.87048309, 0.92274961, 0.00221421, 0.46948837]],

       [[0.98146874, 0.3989448 , 0.81373248, 0.5464565 ],
        [0.77085409, 0.48493107, 0.02911156, 0.08652569],
        [0.11145381, 0.25124511, 0.96491529, 0.63176605]]])

• random.shuffle()

>>> b= np.random.randint(100,200,(3,4))
>>> b
array([[191, 129, 188, 197],
       [192, 179, 170, 135],
       [120, 149, 172, 132]])
>>> np.random.shuffle(b)					# 数组第一轴行随排列，改变原数组
>>> b
array([[120, 149, 172, 132],
       [192, 179, 170, 135],
       [191, 129, 188, 197]])

• random.permutation(x)

>>> b
array([[199, 142, 124, 176],
       [129, 106, 191, 180],
       [117, 118, 177, 156]])
>>> c =np.random.permutation(b)					# 数组第一轴随机排列，乱序数组，不改变原数组
>>> c
array([[129, 106, 191, 180],
       [199, 142, 124, 176],
       [117, 118, 177, 156]])

• random.choice(a[,size,replace,p])

>>> x = np.random.randint(100,200,(8,))
>>> x
array([145, 185, 114, 191, 105, 141, 115, 122])
>>> y=np.random.choice(x,(3,2))					# 从一维数组中以概率p抽取元素，形成size形状的新数组
>>> y
array([[114, 114],
       [122, 105],
       [122, 191]])
>>> y=np.random.choice(x,(3,2),replace=False) 	# replace表示是否可以重用元素，默认为False
>>> y
array([[102, 127],
       [141, 197],
       [129, 141]])
>>> y=np.random.choice(x,(3,2),p=x/np.sum(x))
>>> y
array([[160, 198],
       [198, 150],
       [172, 160]])

统计函数

求和函数sum

>>> a =np.arange(15).reshape(5,3)
>>> a
array([[ 0,  1,  2],
       [ 3,  4,  5],
       [ 6,  7,  8],
       [ 9, 10, 11],
       [12, 13, 14]])
>>> b=np.sum(a,axis=None)			#默认求和所有
>>> b
105

数组期望mean

>>> a
array([[ 0,  1,  2],
       [ 3,  4,  5],
       [ 6,  7,  8],
       [ 9, 10, 11],
       [12, 13, 14]])
>>> c= np.mean(a,axis=1)			#计算数组期望,axis=1时输出第2维度数组的期望
>>> c
array([ 1.,  4.,  7., 10., 13.])

数组加权平均值

>>> d=np.average(a,axis=None,weights=None)
>>> d
7.0
>>> e=np.std(a,axis=None)
>>> e
4.320493798938574

方差var

>>> f=np.var(a,axis=None)
>>> f
18.666666666666668

最大值max，最小值min

>>> x=np.max(a)
>>> y=np.min(a)
>>> x
14
>>> y
0

最小值，最大值降维之后的下标

>>> xx=np.argmax(a)
>>> yy=np.argmin(a)
>>> xx
14
>>> yy
0

一维下标转换多维下标

>>> multiindex = np.unravel_index(1,a.shape)
>>> multiindex
(0, 1)

最大值与最小值的差

>>> cha=np.ptp(a)
>>> cha
14

数组的中位数

>>> mnumber=np.median(a)
>>> mnumber
7.0