通過?pandas.DataFrame.shift命令查看幫助文檔

Signature: pandas.DataFrame.shift(self, periods=1, freq=None, axis=0) Docstring: Shift index by desired number of periods with an optional time freq 

該函數主要的功能就是使數據框中的數據移動，若freq=None時，根據axis的設置，行索引數據保持不變，列索引數據可以在行上上下移動或在列上左右移動；若行索引為時間序列，則可以設置freq參數，根據periods和freq參數值組合，使行索引每次發生periods*freq偏移量滾動，列索引數據不會移動

① 對于DataFrame的行索引是日期型，行索引發生移動，列索引數據不變

In [2]: import pandas as pd  ...: import numpy as np  ...: df = pd.DataFrame(np.arange(24).reshape(6,4),index=pd.date_range(start=  ...: '20170101',periods=6),columns=['A','B','C','D'])  ...: df  ...:Out[2]:       A  B  C  D2017-01-01  0  1  2  32017-01-02  4  5  6  72017-01-03  8  9 10 112017-01-04 12 13 14 152017-01-05 16 17 18 192017-01-06 20 21 22 23In [3]: df.shift(2,axis=0,freq='2D')Out[3]:       A  B  C  D2017-01-05  0  1  2  32017-01-06  4  5  6  72017-01-07  8  9 10 112017-01-08 12 13 14 152017-01-09 16 17 18 192017-01-10 20 21 22 23In [4]: df.shift(2,axis=1,freq='2D')Out[4]:       A  B  C  D2017-01-05  0  1  2  32017-01-06  4  5  6  72017-01-07  8  9 10 112017-01-08 12 13 14 152017-01-09 16 17 18 192017-01-10 20 21 22 23In [5]: df.shift(2,freq='2D')Out[5]:       A  B  C  D2017-01-05  0  1  2  32017-01-06  4  5  6  72017-01-07  8  9 10 112017-01-08 12 13 14 152017-01-09 16 17 18 192017-01-10 20 21 22 23

結論：對于時間索引而言，shift使時間索引發生移動，其他數據保存原樣，且axis設置沒有任何影響

② 對于DataFrame行索引為非時間序列，行索引數據保持不變，列索引數據發生移動

In [6]: import pandas as pd  ...: import numpy as np  ...: df = pd.DataFrame(np.arange(24).reshape(6,4),index=['r1','r2','r3','r4'  ...: ,'r5','r6'],columns=['A','B','C','D'])  ...: df  ...:Out[6]:   A  B  C  Dr1  0  1  2  3r2  4  5  6  7r3  8  9 10 11r4 12 13 14 15r5 16 17 18 19r6 20 21 22 23In [7]: df.shift(periods=2,axis=0)Out[7]:    A   B   C   Dr1  NaN  NaN  NaN  NaNr2  NaN  NaN  NaN  NaNr3  0.0  1.0  2.0  3.0r4  4.0  5.0  6.0  7.0r5  8.0  9.0 10.0 11.0r6 12.0 13.0 14.0 15.0In [8]: df.shift(periods=-2,axis=0)Out[8]:    A   B   C   Dr1  8.0  9.0 10.0 11.0r2 12.0 13.0 14.0 15.0r3 16.0 17.0 18.0 19.0r4 20.0 21.0 22.0 23.0r5  NaN  NaN  NaN  NaNr6  NaN  NaN  NaN  NaNIn [9]: df.shift(periods=2,axis=1)Out[9]:   A  B   C   Dr1 NaN NaN  0.0  1.0r2 NaN NaN  4.0  5.0r3 NaN NaN  8.0  9.0r4 NaN NaN 12.0 13.0r5 NaN NaN 16.0 17.0r6 NaN NaN 20.0 21.0In [10]: df.shift(periods=-2,axis=1)Out[10]:    A   B  C  Dr1  2.0  3.0 NaN NaNr2  6.0  7.0 NaN NaNr3 10.0 11.0 NaN NaNr4 14.0 15.0 NaN NaNr5 18.0 19.0 NaN NaNr6 22.0 23.0 NaN NaN