pandas 笔记：shift

用于将数据系列或数据框中的数据按指定的位置移动。这对于某些时间序列分析特别有用，例如计算数据的变化量或滞后值

1 对Series/DataFrame数据进行移动

1.0 原始数据

import pandas as pd
import numpy as np
df1=pd.DataFrame(np.arange(12).reshape(3,4),
                 columns=['a','b','c','d'],
                 index=['A','B','C'])
df1

'''

    a	b	c	d
A	0	1	2	3
B	4	5	6	7
C	8	9	10	11
'''

1.1  正数：向下移动，负数：向上移动

df1.shift(1)
'''

    a	b	c	d
A	NaN	NaN	NaN	NaN
B	0.0	1.0	2.0	3.0
C	4.0	5.0	6.0	7.0
'''


df1.shift(-2)
'''

    a	b	c	d
A	8.0	9.0	10.011.0
B	NaN	NaN	NaN	NaN
C	NaN	NaN	NaN	NaN
'''

1.2 axis 沿着那个轴shift（默认0）

df1.shift(-2,axis=1)
'''

a	b	c	d
A	2	3	NaN	NaN
B	6	7	NaN	NaN
C	10	11	NaN	NaN
'''

1.3 fill_value (NaN 用什么填充）

df1.shift(-2,
          axis=1,
         fill_value='-1')

'''
	a	b	c	d
A	2	3	-1	-1
B	6	7	-1	-1
C	10	11	-1	-1
'''

1.4 对单列进行shift

df1['a'].shift(-2,
         fill_value='-1')
'''
A     8
B    -1
C    -1
Name: a, dtype: object
'''

2 DateTime shift

import pandas as pd

df=pd.date_range(start='2023-09-13 10:00:00',
             end='2023-09-13 18:00:00',
             freq='1h')
df
'''
DatetimeIndex(['2023-09-13 10:00:00', '2023-09-13 11:00:00',
               '2023-09-13 12:00:00', '2023-09-13 13:00:00',
               '2023-09-13 14:00:00', '2023-09-13 15:00:00',
               '2023-09-13 16:00:00', '2023-09-13 17:00:00',
               '2023-09-13 18:00:00'],
              dtype='datetime64[ns]', freq='H')
'''

默认按照dataframe的frequency 进行shift

df.shift(2,)
'''
DatetimeIndex(['2023-09-13 12:00:00', '2023-09-13 13:00:00',
               '2023-09-13 14:00:00', '2023-09-13 15:00:00',
               '2023-09-13 16:00:00', '2023-09-13 17:00:00',
               '2023-09-13 18:00:00', '2023-09-13 19:00:00',
               '2023-09-13 20:00:00'],
              dtype='datetime64[ns]', freq='H')
'''

freq 手动指定frequency

df.shift(2,freq='1min')
'''
DatetimeIndex(['2023-09-13 10:02:00', '2023-09-13 11:02:00',
               '2023-09-13 12:02:00', '2023-09-13 13:02:00',
               '2023-09-13 14:02:00', '2023-09-13 15:02:00',
               '2023-09-13 16:02:00', '2023-09-13 17:02:00',
               '2023-09-13 18:02:00'],
              dtype='datetime64[ns]', freq='H')
'''