标签：float   match   memory   cas   mem   series   tin   code   may   

Axis in DataFrame

Optional parameter axis may appear in arithmetric between DataFrame and Series,the key point understanding the meaning of axis is match,by default the index of series shall match columns of DataFrame,broadcasting down the rows;And axis may also appear in max(),mean() or so kind of DataFrame object method,by default, axis=‘index‘,meaning find the max one among index,and that is to find the max one of every column.

import pandas as pd

import numpy as np

frame=pd.DataFrame(np.random.randn(4,3),index=[‘Utah‘,‘Ohio‘,‘Texas‘,‘Oregon‘],columns=list(‘bde‘));frame

frame[‘b‘]

Utah     -0.311649
Ohio      0.351583
Texas     0.605527
Oregon    1.577405
Name: b, dtype: float64

series1=frame.iloc[0];series1

b   -0.311649
d    0.252285
e   -0.741715
Name: Utah, dtype: float64

frame.sub(series1,axis=‘columns‘) # By default,arithmetic between DataFrame and Series matches the index of Series on the DataFrame‘s columns,broadcasting down the rows.

frame.sub(series1,axis=1) # The same with above

series2=frame[‘d‘];series2

Utah      0.252285
Ohio      1.287569
Texas    -0.186660
Oregon    0.381833
Name: d, dtype: float64

frame.sub(series2,axis=‘index‘) # Must set axis=‘index‘,so that broadcasts down on column.

frame.max(axis=‘index‘) # max() default to set axis=‘index‘,meaning find the max one among ‘index‘,not every max one of every index.

b    1.577405
d    1.287569
e    1.607757
dtype: float64

Just how to memory this? Exaggeratively imagine that, when DataFrame makes arithmetic operation with Series,the broadcasting direction is across index,downward,by default.And as for max(),mean(),the default set is axis=‘index‘,the finding direction is also across index,downward again. So why both are default to be downward?I do not know, or maybe just because of gravity.HaHa...

Axis in DataFrame

标签：float   match   memory   cas   mem   series   tin   code   may   

原文地址：https://www.cnblogs.com/johnyang/p/12623246.html