LUMO Damage Detection Evaluation Package

This package provides a standardized framework for evaluating damage detection and localization strategies using the LUMO dataset. Users can input timestamps alongside their corresponding anomaly indices, and the package computes various performance scores for each damage case, promoting consistency in damage detection evaluation.

Features

• Standardized Evaluation Metrics: Calculates TPR and FPR at a threshold set such as FPR for training data is 1%. The training dataset should be only the first month of data.
• Damage Case Analysis: Provides detailed performance evaluations for each specific damage scenario within the LUMO dataset.

Installation

To install the package, run:

pip install evaluation_lumo

Usage

To use the package, import the evaluation_lumo.evaluation module and call the compute_tr_by_events function, compute_mean_variation function, or compute_mad function.

from evaluation_lumo.evaluation import compute_tr_by_events, compute_mean_variation, compute_mad

# Example usage

date_index = pd.date_range(start='2021-08-01', ends="2022-08-01", freq='10T')
associated_damage_index = np.random.random(len(date_index))
compute_tr_by_events(date_index, associated_damage_index)
compute_median_variation(date_index, associated_damage_index)
compute_mad(date_index, associated_damage_index)

Contributing

pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change. pre-commit is used to ensure code quality. Please install it before making any changes. run the following command to install pre-commit:

pre-commit run --all-files

then push your changes.