Transformer

TimeSince

Multivariate Missing values Unequal length

Compute element-wise time elapsed between time index and a reference start time.

Creates a column(s) which represents: t - start, where start is a reference time and t is the time index. The type of start must be compatible with the index of X used in .fit() and .transform().

The output can be converted to an integer representing the number of periods elapsed since the start time by setting to_numeric=True. The period is determined by the frequency of the index. For example, if the freq of the index is “MS” or “M” then the output is the integer number of months between t and start.

Quickstart

python

from sktime.transformations.time_since import TimeSince

estimator = TimeSince(start: list [str | datetime | Period | None ] | None=None, *, to_numeric: bool | None=True, freq: str | None=None, keep_original_columns: bool | None=False, positive_only: bool | None=False)

Parameters(5)

starta list of start times, optional, default=None (use earliest time in index)

a “start time” can be one of the following types:

int: Start time to compute the time elapsed, use when index is integer.
time-like: Period or datetime
Start time to compute the time elapsed.

to_numericstring, optional (default=True)

Return the integer number of periods elapsed since start; the period is defined by the frequency of the data. Converts datetime types to pd.Period before calculating time differences.

freq‘str’, optional, default=None

Only used when X has a pd.DatetimeIndex without a specified frequency. Specifies the frequency of the index of your data. The string should match a pandas offset alias:

https://pandas.pydata.org/pandas-docs/stable/user_guide/timeseries.html#offset-aliases

keep_original_columnsboolean, optional, default=False

Keep original columns in X passed to .transform().

positive_onlyboolean, optional, default=False

Clips negative values to zero when to_numeric is True.

Examples

>>> from sktime.datasets import load_airline
>>> from sktime.transformations.time_since import TimeSince
>>> X = load_airline () Create a single column with time elapsed since start date of time series. The output is in units of integer number of months, same as the index freq.
>>> transformer = TimeSince ()
>>> Xt = transformer. fit_transform (X) Create multiple columns with different start times. The output is in units of integer number of months, same as the index freq.
>>> transformer = TimeSince (["2000-01", "2000-02" ])
>>> Xt = transformer. fit_transform (X)