vastorbit.machine_learning.vast.tsa.VAR.predict¶
- VAR.predict(vdf: Annotated[str | VastFrame, ''] | None = None, ts: str | None = None, y: Annotated[str | list[str], 'STRING representing one column or a list of columns'] | None = None, start: int | None = None, npredictions: int = 10, output_standard_errors: bool = False, output_index: bool = False, output_estimated_ts: bool = False, freq: Literal[None, 'm', 'months', 'y', 'year', 'infer'] = 'infer', filter_step: int | None = None, method: Literal['auto', 'forecast'] = 'auto', use_index_as_suffix: bool = False) VastFrame¶
Predicts using the input relation.
For SQL-trained AR/VAR models, this generates predictions using the learned coefficients directly in SQL.
- Parameters:
vdf (SQLRelation, optional) – Object used to run the prediction.
ts (str, optional) – TS column used to order the data.
y (SQLColumns, optional) – Response column(s).
start (int, optional) – Starting position for prediction.
npredictions (int, optional) – Number of predicted timesteps.
output_standard_errors (bool, optional) – Whether to return standard errors (not supported for SQL models).
output_index (bool, optional) – Whether to return the index.
output_estimated_ts (bool, optional) – Whether to estimate timestamps.
freq (str, optional) – Frequency for timestamp estimation.
filter_step (int, optional) – Filter frequency for predictions.
method (str, optional) – Forecasting method (‘auto’ or ‘forecast’).
use_index_as_suffix (bool, optional) – For multivariate models.
- Returns:
Predictions.
- Return type:
- Raises:
NotImplementedError – For ARIMA models.