Random Forest model implementation for time series forecasting

RandomForestHyperparams

random_forest_model.RandomForestHyperparams · inherits PydanticBaseModel

Attribute	Type	Default	Description
n_estimators	int	-	Number of trees in the forest
max_depth	int	-	Maximum depth of trees
min_samples_split	int	2	Minimum samples to split a node
min_samples_leaf	int	1	Minimum samples in a leaf
max_features	Literal["sqrt", "log2", "auto"]	"sqrt"	Number of features to consider for splits

RandomForestModel

random_forest_model.RandomForestModel · inherits BaseModel

__init__(self, params: Dict[str, Any], settings: Dict[str, Any])

Initialize Random Forest model with model-specific parameters.

Parameter	Type	Default	Description
params	Dict[str, Any]	-	(undocumented)
settings	Dict[str, Any]	-	(undocumented)

train(self, y_context: numpy.ndarray, y_target: numpy.ndarray, timestamps_context: numpy.ndarray, timestamps_target: numpy.ndarray, x_context: Optional[numpy.ndarray] = None, x_target: Optional[numpy.ndarray] = None, **kwargs: dict) -> RandomForestModel

Train the Random Forest model on given data.

TECHNIQUE: Single Model with Timestamp Features - Creates lag features from historical target values - Adds rolling statistics (mean, std, min, max) - Includes trend features using linear regression - Incorporates timestamp features for time-aware splits - Uses a single model to predict all forecast horizon steps

Parameter	Type	Default	Description
y_context	numpy.ndarray	-	Past target values (time series) - used for training
y_target	numpy.ndarray	-	Future target values (optional, for validation)
timestamps_context	numpy.ndarray	-	Timestamps for y_context (optional)
timestamps_target	numpy.ndarray	-	Timestamps for y_target (optional)
x_context	Optional[numpy.ndarray]	None	(undocumented)
x_target	Optional[numpy.ndarray]	None	(undocumented)
**kwargs	dict	-	Additional keyword arguments

Returns: RandomForestModel (The fitted model instance.)

predict(self, y_context: numpy.ndarray, timestamps_context: numpy.ndarray, timestamps_target: numpy.ndarray, x_context: Optional[numpy.ndarray] = None, x_target: Optional[numpy.ndarray] = None, **kwargs: dict) -> numpy.ndarray

Make predictions using the trained Random Forest model.

TECHNIQUE: Single Model with Timestamp Features - Uses last lookback_window values to create features - Incorporates timestamp features for time-aware prediction - Predicts all forecast horizon steps with a single model

Parameter	Type	Default	Description
y_context	numpy.ndarray	-	Past target values (time series) - used for prediction
timestamps_context	numpy.ndarray	-	Timestamps for y_context (optional)
timestamps_target	numpy.ndarray	-	Timestamps for y_target (optional)
x_context	Optional[numpy.ndarray]	None	(undocumented)
x_target	Optional[numpy.ndarray]	None	(undocumented)
**kwargs	dict	-	Additional keyword arguments

Returns: numpy.ndarray (Model predictions with shape (1, forecast_horizon)) Raises: ValueError (Model is not trained yet. Call train() first.)