Library for studying the performance of data reconstruction models developed for (but not limited to) machine learning reconstruction of events from particle physics experiments.
from RPP.plotters.classification_plotter import ClassificationPlotter
# Initialize plotter
FlavourPlotter = ClassificationPlotter('/my/plot/dir', 'target_label')
# Add model predictions
FlavourPlotter.add_results('my_predictions.csv', 'my_data.db', 'ModelName')
# Plot model score histogram and performance (ROC) curve
FlavourPlotter.plot_score_hist()
FlavourPlotter.plot_performance_curve()The current data structure is based on the approach of the GraphNeT framework with model predictions as .csv-files and metadata as sqlite-databases.