multitask.strategies.TrainingStrategy

multitask.strategies.TrainingStrategy()

Strategy interface for preparing a forecaster before the final fit.

Implementations return a forecaster with any tuning/parameter changes applied. The final forecaster.fit(...) and prediction packaging are performed by BaseTask._train_and_predict_target after this call.

Examples

import pandas as pd
from spotforecast2_safe.multitask.strategies import (
    TrainingStrategy,
    LazyStrategy,
    DefaultsStrategy,
)

# Both concrete strategies satisfy the TrainingStrategy protocol:
# they expose a `name` attribute and a `prepare_forecaster` method.
for cls in (LazyStrategy, DefaultsStrategy):
    s = cls()
    assert hasattr(s, "name"), f"{cls.__name__} missing .name"
    assert callable(s.prepare_forecaster), f"{cls.__name__} missing .prepare_forecaster"
    print(f"{cls.__name__}.name = {s.name!r}")

LazyStrategy.name = 'lazy'
DefaultsStrategy.name = 'defaults'

Methods

Name	Description
prepare_forecaster	Return a forecaster ready for the final fit step.

prepare_forecaster

multitask.strategies.TrainingStrategy.prepare_forecaster(
    task,
    target,
    forecaster,
    y_train,
    exog_train=None,
)

Return a forecaster ready for the final fit step.

Examples

import pandas as pd
from sklearn.linear_model import LinearRegression
from spotforecast2_safe.forecaster.recursive import ForecasterRecursive
from spotforecast2_safe.multitask.strategies import DefaultsStrategy

# Demonstrate prepare_forecaster via a concrete implementation.
# DefaultsStrategy is the simplest: it returns the forecaster unchanged.
forecaster = ForecasterRecursive(estimator=LinearRegression(), lags=3)
y_train = pd.Series(range(30), dtype=float, name="target_0")

class _NullTask:
    pass

strategy = DefaultsStrategy()
result = strategy.prepare_forecaster(_NullTask(), "target_0", forecaster, y_train)
assert result is forecaster
print(f"prepare_forecaster returned the same object: {result is forecaster}")

prepare_forecaster returned the same object: True