multitask.predict.execute_predict

multitask.predict.execute_predict(
    task,
    show=False,
    task_name=None,
    max_age_days=None,
)

Execute prediction-only mode using previously saved models.

Loads the most recent fitted forecaster for every configured target from the cache directory. No training is performed. If no saved models can be found the function raises RuntimeError.

Parameters

Name	Type	Description	Default
task	BaseTask	A `BaseTask` (or subclass) instance with prepared data.	required
show	bool	If `True`, invoke the visualisation hooks.	`False`
task_name	Optional[str]	Restrict model loading to a specific source task (`"lazy"`, `"defaults"`, `"optuna"`, or `"spotoptim"`). `None` loads the most recent model regardless of source.	`None`
max_age_days	Optional[float]	Maximum age in days for saved models. Models older than this are ignored. `None` accepts any age.	`None`

Returns

Name	Type	Description
	Dict[str, Any]	Aggregated prediction package (weighted combination of all targets).
	Dict[str, Any]	Per-target packages are stored on `task.results["predict"]`.

Raises

Name	Type	Description
	RuntimeError	If no saved models are found in the cache directory, or if a target has no matching saved model.

Examples

import tempfile
import warnings
from pathlib import Path

import numpy as np
import pandas as pd

from spotforecast2_safe.configurator.config_multi import ConfigMulti
from spotforecast2_safe.multitask.lazy import LazyTask
from spotforecast2_safe.multitask.predict import PredictTask, execute_predict

warnings.filterwarnings("ignore")

rng = np.random.default_rng(0)
idx = pd.date_range("2023-01-01", periods=300, freq="h", tz="UTC")
df = pd.DataFrame(
    {"total_load_actual": rng.normal(10_000, 500, 300).clip(8_000, 12_000)},
    index=idx,
)
df.index.name = "DateTime"

with tempfile.TemporaryDirectory() as tmp:
    cfg = ConfigMulti(
        data_frame_name="demo",
        predict_size=12,
        auto_save_models=True,
        cache_home=Path(tmp),
        targets=["total_load_actual"],
        verbose=False,
        use_outlier_detection=False,
        use_exogenous_features=False,
        number_folds=2,
    )
    # Train and save a model with LazyTask first.
    lazy = LazyTask(cfg, dataframe=df)
    lazy.prepare_data()
    lazy.run()

    # Load the saved model and generate predictions.
    predict_task = PredictTask(cfg, dataframe=df)
    predict_task.prepare_data()
    result = execute_predict(predict_task)

    print(f"Future predictions shape: {result['future_pred'].shape}")
    assert result["future_pred"].shape == (12,)

Future predictions shape: (12,)