Introduction¶

# define model
regressor = RandomForestRegressor()

# transform to tabular form
features_no_of_lags = 3 # {1, 2, 3}
forecaster = make_reduction(
    regressor,
    window_length = features_no_of_lags,
    strategy = "recursive"
)

forecast_horizon = np.arange(1, 5)

cv_expanding = ExpandingWindowSplitter(
    initial_window = 24*10,
    step_length = 24,
    fh = forecast_horizon,
)

cv_expanding = SlidingWindowSplitter(
    window_length = 24*10,
    step_length = 24,
    fh = forecast_horizon,
)

results = evaluate(
    forecaster = forecaster,
    y = y_train,
    cv = cv,
    return_data = True,
    strategy = "refit" # ["refit", "update"]
)