Data types

Contents

Data types#

pandas generally inherits its types system from NumPy. Below, we consider some features exclusive to pandas, and options for processing pandas series/data frames specific to particular data types.

Date & time#

Pandas provides an option for specifying a time zone for a datetime object.

The following table lists features of the Pandas design that allow you to work with datetime and timedelta objects.

Function / Method	Description
`pd.to_datetime()`	Convert strings, numbers, etc. to `datetime64[ns]`
`pd.date_range()`	Create a range of dates with fixed frequency
`pd.bdate_range()`	Like `date_range` but excludes weekends/holidays (business days)
`pd.to_timedelta()`	Convert strings or numbers to `timedelta64[ns]`
`pd.timedelta_range()`	Create a range of timedeltas
`Series.dt`	Accessor for datetime and timedelta components (e.g. `.dt.year`, `.dt.days`)
`Series.dt.date`	Extract just the date part (as `datetime.date`)
`Series.dt.time`	Extract just the time part (as `datetime.time`)
`Series.dt.tz_localize()`	Localize naive datetime to a timezone
`Series.dt.tz_convert()`	Convert timezone-aware datetimes to another timezone
`Series.dt.normalize()`	Set time to midnight (00:00:00)
`Series.dt.floor(freq)`	Round down to nearest frequency
`Series.dt.ceil(freq)`	Round up to nearest frequency
`Series.dt.round(freq)`	Round to nearest frequency
`Series.dt.strftime(fmt)`	Format datetime as string using `strftime` format
`Series.dt.total_seconds()`	Get total seconds from timedelta
`Series.dt.days` / `dt.seconds`	Extract days or seconds from timedelta
`DataFrame.resample()`	Resample time series data
`DataFrame.asfreq(freq)`	Change frequency without resampling
`DataFrame.shift(periods, freq)`	Shift data in time
`DataFrame.diff()`	Calculate timedelta difference between rows
`DatetimeIndex`, `TimedeltaIndex`	Specialized index types for time-aware indexing and slicing
`DataFrame.set_index("timestamp")`	Useful to enable datetime indexing/resampling
`DataFrame.rolling(window="30D")`	Rolling operations over time-based windows

Check details in: