The amount to shift by. If axis=0 (the default), then positive integers mean rows are shifted downwards, while negative integers mean rows are shifted upwards. The same logic applies for when axis=1 (column shifts).

By default, periods=1.

2. freqlink | DateOffset or timedelta or string | optional

The time unit to shift by. By default, freq=None.

3. axis | int or string | optional

Whether to shift the row index or the column index:

Axis	Description
`0` or `"index"`	Shift row index.
`1` or `"columns"`	Shift column index.

By default, axis=0.

Return Value

A new DataFrame with a shifted index.

Examples

Consider the following DataFrame:


        
        
            
                
                
                    date_index = pd.date_range("2020/12/25", periods=4)
df = pd.DataFrame({"A":[2,3,4,5],"B":[6,7,8,9]}, index=date_index)
df
                
            
                         A   B
2020-12-25   2   6
2020-12-26   3   7
2020-12-27   4   8
2020-12-28   5   9

Here, the index of df is of type DatetimeIndex.

Periods parameter

To shift the row index by 1 day:


        
        
            
                
                
                    df.tshift()
                
            
                         A   B
2020-12-26   2   6
2020-12-27   3   7
2020-12-28   4   8
2020-12-29   5   9

To shift the row index by 2 days:


        
        
            
                
                
                    df.tshift(periods=2)
                
            
                         A   B
2020-12-27   2   6
2020-12-28   3   7
2020-12-29   4   8
2020-12-30   5   9

To back-shift the row index by 1 day, set a negative period like so:


        
        
            
                
                
                    df.tshift(periods=-1)
                
            
                         A   B
2020-12-24   2   6
2020-12-25   3   7
2020-12-26   4   8
2020-12-27   5   9

Frequency parameter

Consider the same DataFrame as above:


        
        
            
                
                
                    df
                
            
                         A   B
2020-12-25   2   6
2020-12-26   3   7
2020-12-27   4   8
2020-12-28   5   9

By default, freq is inferred from the date index. Since our df has a freq of a day, that is, each date index varies by a single day, tshift will use freq="D":


        
        
            
                
                
                    df.tshift(periods=1)    # freq="D"
                
            
                         A   B
2020-12-26   2   6
2020-12-27   3   7
2020-12-28   4   8
2020-12-29   5   9

Here, each date index is shifted by 1 day unit. To shift by 3 day units, for instance:


        
        
            
                
                
                    df.tshift(periods=1, freq="3D")
                
            
                         A   B
2020-12-28   2   6
2020-12-29   3   7
2020-12-30   4   8
2020-12-31   5   9

Note that this essentially has the same effect as specifying periods=3:


        
        
            
                
                
                    df.tshift(periods=3)
                
            
                         A   B
2020-12-28   2   6
2020-12-29   3   7
2020-12-30   4   8
2020-12-31   5   9

Warning

Consider the following DataFrame:


        
        
            
                
                
                    date_index = pd.date_range("2020/12/30", periods=4)
df = pd.DataFrame({"A":[2,3,4,5],"B":[6,7,8,9]}, index=date_index)
df
                
            
                         A   B
2020-12-30   2   6
2020-12-31   3   7
2021-01-01   4   8
2021-01-02   5   9

Setting freq="M" yields:


        
        
            
                
                
                    df.tshift(periods=1, freq="M")
                
            
                         A   B
2020-12-31   2   6
2021-01-31   3   7
2021-01-31   4   8
2021-01-31   5   9

The results is slightly unintuitive since we'd expect an one-month offset to be applied to each date index.

Note the following:

each date index got shifted to the end of the month.
since we specified freq="M", the day unit of the date index becomes irrelevant.
the exception is when the day unit is at the end of the month (e.g. 12-31), in which case the month will get shifted (e.g. to 01-31).

Pandas DataFrame | shift method

Shifts the rows or columns by the specified amount, without changing the size of the DataFrame.

chevron_right