numpy和pandas axis的差异

Posted 2020-11-21 yitiaodahe

1.numpy

arr = np.random.randn(5,4)#正态分布数据
print(arr)
print(arr.sum()) # 数组/矩阵中所有元素求和,等价于np.sum(arr)
print(np.sum(arr, axis=0))# 按列去求和; 
print(arr.sum(axis=1))  # 按行去求和;

[[ 0.04154798  0.25697028  2.36239272 -1.72886735]
 [ 0.50448843 -0.63285194  2.9090727   0.61004107]
 [ 0.10730241 -0.13162546 -0.67925053  0.12864452]
 [ 0.04125252 -0.03968486 -0.60453958  0.94637586]
 [ 1.65060502 -0.18266035 -1.06259085  0.18515147]]
4.681774035077944
[ 2.34519635 -0.72985234  2.92508445  0.14134557]
[ 0.93204362  3.39075026 -0.57492907  0.34340393  0.59050528]
2.pandas　　

df

	id	date	city	category	age	price
one	1001	2013-01-02	Beijing	100-A	23	1200.0
two	1002	2013-01-03	NaN	100-B	44	NaN
three	1003	2013-01-04	guangzhou	110-A	54	2133.0
four	1004	2013-01-05	Shenzhen	110-C	32	5433.0
five	1005	2013-01-06	shanghai	210-A	34	NaN
six	1006	2013-01-07	BEIJING	130-F	32	4432.0

 

 

 

 

 

 

df.dropna(axis=1)#针对列向有nan值的情况

	id	date	category	age
one	1001	2013-01-02	100-A	23
two	1002	2013-01-03	100-B	44
three	1003	2013-01-04	110-A	54
four	1004	2013-01-05	110-C	32
five	1005	2013-01-06	210-A	34
six	1006	2013-01-07	130-F	32