pandas.Series.dt.shift#

Series.dt.shift(periods=1, freq=None)[source]#

Shift the values by desired number of time frequency increments.

Unlike Series.shift(), which moves values to different positions, this adds periods * freq to each value, like DatetimeIndex.shift().

Parameters:

periodsint, default 1: Number of periods (or increments) to shift by, can be positive or negative.
freqpandas.DateOffset, pandas.Timedelta or str, optional: Frequency increment to shift by. If None, the frequency inferred from the values (see Series.dt.freq) is used, so values in descending order move backward for positive periods. Must be None for period values, which are shifted by their own frequency.

Returns:

Series: Series with the shifted values, indexed like the original Series.

Raises:

NullFrequencyError: If freq is None, periods is nonzero, and no frequency can be inferred from the values.
TypeError: If freq is passed for period values.

See also

DatetimeIndex.shift: Shift values of DatetimeIndex.
Series.shift: Shift values by desired number of positions.

Examples

>>> ser = pd.Series(pd.date_range("2020-01-01", periods=3, freq="D"))
>>> ser.dt.shift(1)
0   2020-01-02
1   2020-01-03
2   2020-01-04
dtype: datetime64[us]

>>> ser.dt.shift(2, freq="h")
0   2020-01-01 02:00:00
1   2020-01-02 02:00:00
2   2020-01-03 02:00:00
dtype: datetime64[us]