ResultsGenerator

results_generator.ResultsGenerator

Class to generate pivot tables from evaluations CSV and compute aggregations.

Automatically detects all subclasses of BaseAggregator and initializes them for computing aggregation scores.

__init__(self, pivot_tables: Optional[Dict[str, pd.DataFrame]] = None, baseline_model: str = 'seasonal_naive')

Initialize the ResultsGenerator.

Parameter	Type	Default	Description
pivot_tables	Optional[Dict[str, pd.DataFrame]]	None	Optional dictionary mapping metric names to pivot tables. Each pivot table has models as index, tasks as columns, scores as values. If None, must be set later using set_pivot_tables.
baseline_model	str	'seasonal_naive'	Name of the baseline model for aggregators that need it (default: seasonal_naive)

set_pivot_tables(self, pivot_tables: Dict[str, pd.DataFrame])

Set the pivot tables dictionary.

Parameters:

Parameter	Type	Default	Description
pivot_tables	Dict[str, pd.DataFrame]	-	Dictionary mapping metric names to pivot tables

create_pivot_tables(csv_path: str) -> Dict[str, pd.DataFrame]

Read evaluations CSV, drop best_params column, and create pivot tables for each metric column.

Parameters:

Parameter	Type	Default	Description
csv_path	str	-	Path to the evaluations CSV file

Returns: Dict[str, pd.DataFrame] (Dictionary mapping metric names to pivot tables).

compute_aggregations(self, metric_name: Optional[str] = None) -> Dict[str, Dict[str, pd.Series]]

Compute all aggregation scores for pivot tables.

Parameters:

Parameter	Type	Default	Description
metric_name	Optional[str]	None	Optional metric name to compute aggregations for. If None, computes for all pivot tables.

Returns: Dict[str, Dict[str, pd.Series]] (Dictionary mapping metric names to dictionaries of aggregator results. Format: {metric_name: {aggregator_name: Series}}).

save_pivot_tables(self, output_dir: str)

Save all pivot tables to CSV files.

Parameters:

Parameter	Type	Default	Description
output_dir	str	-	Directory to save pivot tables

save_aggregations(self, output_dir: str)

Compute aggregations for all pivot tables and save results to CSV files.

Parameters:

Parameter	Type	Default	Description
output_dir	str	-	Directory to save aggregation results