import numpy as np
[docs]
def timedelta_to_nb_observations(timedelta_value: np.timedelta64, time_stamps: np.array) -> int:
"""
A utility method to convert time delta values to a number of observations. This method
will search for the minimal number of observations such that they cover at least the
time period given by ``timedelta_value`` in the provided ``time_stamps``. This method
assumes that the provided ``time_stamps`` are sorted, with the earliest time stamp
at the first position and the latest time stamp at the last position. The given
time stamps should be regularly sampled.
Parameters
----------
timedelta_value: np.timedelta64
The time interval that must be covered by the computed number of observations.
time_stamps: np.array of np.datetime64
The time stamps corresponding to the observation times of a time series.
Returns
-------
nb_observations: int
The minimal number of consecutive observations that cover the given timedelta
value within the given time stamps.
"""
if time_stamps[-1] - time_stamps[0] < timedelta_value:
raise ValueError("The given timedelta value is too large for the given time stamps!")
nb_observations = 1
while time_stamps[nb_observations] - time_stamps[0] < timedelta_value:
nb_observations += 1
return nb_observations