Additional util functions¶

class darts.utils.ts_utils.SeriesType(value)[source]¶

Bases: Enum

An Enum for different TimeSeries sequence types.

NONE = -1¶

SEQ = 1¶

SEQ_SEQ = 2¶

SINGLE = 0¶

darts.utils.ts_utils.get_series_seq_type(ts)[source]¶

Returns the sequence type of ts.

SeriesType.SINGLE: TimeSeries (e.g. a single series)
SeriesType.SEQ: sequence of TimeSeries (e.g. multiple series)
SeriesType.SEQ_SEQ: sequence of sequences of TimeSeries (e.g. historical forecasts output)

Parameters: ts (Union[TimeSeries, Sequence[TimeSeries], Sequence[Sequence[TimeSeries]]]) – The input series to get the sequence type from.
Raises: ValueError – If ts does not have one of the expected sequence types.
Return type: SeriesType

darts.utils.ts_utils.get_single_series(ts)[source]¶

Returns a single (first) TimeSeries or None from ts. Returns ts if ts is a TimeSeries, ts[0] if ts is a Sequence[TimeSeries], and ts[0][0] if ts is a Sequence[Sequence[TimeSeries]]. Otherwise, returns None.

Parameters: ts (Union[TimeSeries, Sequence[TimeSeries], Sequence[Sequence[TimeSeries]], None]) – None, a single TimeSeries, a sequence of TimeSeries, or a sequence of sequences of TimeSeries.
Returns: ts if ts is a TimeSeries, ts[0] if ts is a Sequence of TimeSeries. Otherwise, returns None
Return type: TimeSeries

darts.utils.ts_utils.retain_period_common_to_all(series)[source]¶

Trims all series in the provided list, if necessary, so that the returned time series have a common span (corresponding to largest time sub-interval common to all series).

Parameters: series (List[TimeSeries]) – The list of series to consider.
Raises: ValueError – If no common time sub-interval exists
Returns: A list of series, where each series have the same span
Return type: List[TimeSeries]

darts.utils.ts_utils.seq2series(ts)[source]¶

If ts is a Sequence with only a single series, return the single series as TimeSeries.

Parameters: ts (Union[TimeSeries, Sequence[TimeSeries], None]) – None, a single TimeSeries, or a sequence of TimeSeries
Return type: ts if ts if is not a single element TimeSeries sequence, else ts[0]

darts.utils.ts_utils.series2seq(ts, seq_type_out=SeriesType.SEQ, nested=False)[source]¶

If possible, converts ts into the desired sequence type seq_type_out. Otherwise, returns the original ts.

Parameters

ts (Union[TimeSeries, Sequence[TimeSeries], Sequence[Sequence[TimeSeries]], None]) – None, a single TimeSeries, a sequence of TimeSeries, or a sequence of sequences of TimeSeries.
seq_type_out (SeriesType) –
The output sequence type:
- SeriesType.SINGLE: TimeSeries (e.g. a single series)
- SeriesType.SEQ: sequence of TimeSeries (e.g. multiple series)
- SeriesType.SEQ_SEQ: sequence of sequences of TimeSeries (e.g. historical forecasts output)
nested (bool) – Only applies with seq_type_out=SeriesType.SEQ_SEQ and ts having a sequence type SeriesType.SEQ. In this case, wrap each element in ts in a list ([ts1, ts2] -> [[ts1], [ts2]]).

Raises

ValueError – If there is an invalid seq_type_out value.

Return type

Union[TimeSeries, Sequence[TimeSeries], Sequence[Sequence[TimeSeries]], None]

Utils for Pytorch and its usage

Additional util functions