Saving results

The output can be stored either in a single JSON file or in multiple JSON files. Each JSON file contains an array of objects.

Output

The save_json function is responsible for saving the results. A dedicated directory named results_yyyymmdd_hhmmss is automatically created to store all generated files.

scentree.io.writer.save_json

save_json(output_dir: Union[str, Path], num_stages: int, predicted_value: List[NDArray], observed_value: Union[List[NDArray], None], scenario_trees: ScenarioTrees, mapping_datasets_columns: DatasetMappings, in_sample_prediction: bool = True, multiple_files: bool = False) -> None

Save model results and scenario trees into a JSON file.

This function stores simulation results, predictions, and tree structures into a timestamped output directory. It supports both single-file and multi-file export modes.

Parameters:

Name	Type	Description	Default
output_dir	Union[str, Path]	Directory where results will be saved.	required
num_stages	int	Number of stages in the scenario model.	required
predicted_value	List[NDArray]	List of predicted value arrays, one per tree.	required
observed_value	Union[List[NDArray], None]	List of observed values, or None if not available.	required
scenario_trees	ScenarioTrees	List of scenario tree. Each position contains a scenario tree.	required
mapping_datasets_columns	DatasetMappings	Mapping between dataset names, their column indices and their stages.	required
in_sample_prediction	bool	Whether predictions are in-sample. Default to true.	True
multiple_files	bool	If True, results are saved in separate files (not yet implemented). If False, all results are stored in a single JSON file.	False

Raises:

Type	Description
FileNotFoundError	If output_dir does not exist.
NotADirectoryError	If output_dir is not a directory.

Description of the file

The JSON file contains an array of objects. Each object includes the following fields:

Field	Type	Description
num_scenarios	integer	Number of generated scenarios.
num_stages	integer	Number of stages in the stochastic problem.
in_sample_prediction	boolean	Indicates whether the generated scenarios correspond to observed days (true) or future days (false).
scenario_tree_data	array of array of float	Scenario data. It contains one array per scenario, where each inner array represents a single scenario.
mean_value_scenario_tree	array of float	Mean value of the generated scenarios.
predicted_value	array of float	Predicted values.
observed_value	array of float or null	Observed values. If in_sample_prediction is true, this field contains an array of length data_dim; otherwise, it is null.
scenario_probabilities	array of float	Probability associated with each generated scenario.
mapping_datasets_columns	array of object	Mapping between the original datasets and the corresponding columns in the scenario data. Each object contains the dataset name and the associated column indices.
tree	array of object	Description of the generated scenario tree. The structure of each object is described in the table below.

Tree object

Field	Type	Description
key	int[2]	A pair representing (stage, position in stage).
scenario_ids	int[]	Array of scenario IDs that belong to the same cluster.
parent_key	int[2] or null	Parent key of the current node. It is null for root nodes.
description	string	Textual description of the key field.