Data limits expansion#

Data limits expansion allows bounds inferred from and available data argument to stretch outward slightly, toward the likely “real” limits of the data.

from pandas import Timestamp
from time_split.support import expand_limits

limits = Timestamp("2019-05-11"), Timestamp("2019-05-11 22:05:30")

expanded = expand_limits(limits, "d")
assert expanded == (
    Timestamp('2019-05-11 00:00:00'),
    Timestamp('2019-05-12 00:00:00'),
)

See support.expand_limits() for more examples and manual experimentation.

`ExpandLimits` specification options.#
Type	Description
`False`	Disable expand_limits; use real limits instead.
`True` or `'auto'`	Auto-expand_limits using `settings.auto_expand_limits`-settings. Snap limits to the nearest `hour` or `day`, depending on the amount of available data. Use `settings.auto_expand_limits.set_level` to modify auto-expand_limits behavior.
`str`	Manual expand_limits specification. Pass an offset alias specify how limits should be rounded. To specify by how much limits may be rounded, pass two offset aliases separated by a `'<'`. For example, passing `expand_limits="d<1h"` will snap limits to the nearest date, but will not expand limits by more than one hour in either direction.

The expand_limits() function uses level tuples on the form (start_at, round_to, tolerance) internally. These may be passed directly to expand_limits(), but not to any other functions. Use the types above instead.