pandas.Interval — pandas 3.0.0.dev0+2096.g8045c2d59f documentation (original) (raw)
class pandas.Interval#
Immutable object implementing an Interval, a bounded slice-like interval.
Attributes
See also
An Index of Interval objects that are all closed on the same side.
Convert continuous data into discrete bins (Categorical of Interval objects).
Convert continuous data into bins (Categorical of Interval objects) based on quantiles.
Represents a period of time.
Notes
The parameters left and right must be from the same type, you must be able to compare them and they must satisfy left <= right.
A closed interval (in mathematics denoted by square brackets) contains its endpoints, i.e. the closed interval [0, 5] is characterized by the conditions 0 <= x <= 5. This is what closed='both' stands for. An open interval (in mathematics denoted by parentheses) does not contain its endpoints, i.e. the open interval (0, 5) is characterized by the conditions 0 < x < 5. This is what closed='neither' stands for. Intervals can also be half-open or half-closed, i.e. [0, 5) is described by 0 <= x < 5 (closed='left') and (0, 5] is described by 0 < x <= 5 (closed='right').
Examples
It is possible to build Intervals of different types, like numeric ones:
iv = pd.Interval(left=0, right=5) iv Interval(0, 5, closed='right')
You can check if an element belongs to it, or if it contains another interval:
2.5 in iv True pd.Interval(left=2, right=5, closed='both') in iv True
You can test the bounds (closed='right', so 0 < x <= 5):
0 in iv False 5 in iv True 0.0001 in iv True
Calculate its length
You can operate with + and * over an Interval and the operation is applied to each of its bounds, so the result depends on the type of the bound elements
shifted_iv = iv + 3 shifted_iv Interval(3, 8, closed='right') extended_iv = iv * 10.0 extended_iv Interval(0.0, 50.0, closed='right')
To create a time interval you can use Timestamps as the bounds
year_2017 = pd.Interval(pd.Timestamp('2017-01-01 00:00:00'), ... pd.Timestamp('2018-01-01 00:00:00'), ... closed='left') pd.Timestamp('2017-01-01 00:00') in year_2017 True year_2017.length Timedelta('365 days 00:00:00')
Attributes
Methods