Endperformance Analysis module

endperformance_analysis.baseline(random_sampling: str, prediction_entropy_uncertainty: str, prediction_entropy: str, filter_strategies: str) → None

Calculate and compare baseline metrics for different filtering strategies.

This function loads data for different sampling and filtering strategies, calculates rankings for each strategy across tasks, and prints the final rankings.

Args:

random_sampling (str): Directory name for random sampling assets prediction_entropy_uncertainty (str): Directory name for prediction entropy uncertainty assets prediction_entropy (str): Directory name for prediction entropy assets filter_strategies (str): Directory name for filter strategies assets

Returns:

None: Results are printed to standard output

endperformance_analysis.collect_asset_paths(ASSET_PATHS: Path) → DefaultDict[str, List]

Collect paths to all asset files ending with ‘_f1s.npy’ from a directory.

This function traverses the given directory to find files ending with ‘_f1s.npy’ and organizes them by task name.

Args:

ASSET_PATHS (Path): Directory path to search for asset files

Returns:

DefaultDict[str, List]: A dictionary mapping task names to lists of file paths

endperformance_analysis.load_asset_data(workspace_data: DefaultDict[str, List[Path]]) → Dict[str, DataFrame]

Load and organize asset data from file paths into pandas DataFrames.

This function processes a collection of file paths, loads the numpy arrays, and organizes them by filter strategy name into DataFrames.

Args:

workspace_data (DefaultDict[str, List[Path]]): Dictionary mapping task names

to lists of file paths

Returns:

Dict[str, pd.DataFrame]: Dictionary mapping task names to DataFrames containing

the processed data

endperformance_analysis.load_asset_data_baseline(ASSET_PATHS: Path) → DefaultDict[str, List]

Load baseline asset data from a directory, calculating mean F1 scores.

This function traverses the given directory to find files ending with ‘f1s.npy’, loads their content, and calculates the mean value for each task.

Args:

ASSET_PATHS (Path): Directory path containing the asset files

Returns:

DefaultDict[str, List]: A dictionary mapping task names to lists of mean F1 scores

endperformance_analysis.transform_into_experimental_data(data)

Transform the data structure for experimental analysis.

This function explodes the DataFrames and concatenates them into a single DataFrame for further analysis.

Args:

data (Dict[str, pd.DataFrame]): Dictionary of DataFrames to transform

Returns:

pd.DataFrame: A single concatenated DataFrame with exploded series