Save analysis results#
After running miniML analysis, the results can be saved to a file. miniML currently supports saving the results to text (.csv), Pickle (.pickle) and HDF5 (.h5) files.
Saving to file#
Clicking the Save button opens a choose file dialog. The default name is the name of the recording. The file can be saved as a .csv, .pickle or .h5 file.
CSV#
When saving as CSV, the average results are saved in the *_avg.csv file and the individual event statistics in the *_individual.csv file. Individual event statistics include event location (in samples), model prediction value (“score”), amplitude, charge, risetime and decay time. The average values saved are mean event amplitude (in data units), standard deviation of amplitudes (in data units), median event amplitude (in data units), mean charge (in data units x s), mean 10-90 risetime (in s), mean decay time (in s), mean halfwidth (in s), decay time constant from an exponential fit to the average event waveform (in s), and event frequency (in Hz).
HDF5#
When saving as HDF5, the results are saved in a single HDF5 file. The file contains the following datasets:
events: all detected events
event_params: event_locations, event_scores, event_amplitudes, event_charges, event_risetimes, event_halfdecays, event_bsls
event_statistics: amplitude_average, amplitude_stdev, amplitude_median, charge_mean, charge_median, risetime_mean, risetime_median, decaytime_mean, decaytime_median, halfwidth_mean, halfwidth_median, decay_from_fit, frequency The file contains the following attributes: amplitude_unit, recording_time, source_filename, miniml_model, miniml_model_threshold, stride, window, event_direction, filter_factor, gradient_convolve_win, relative_prominence, deleted_events
PICKLE#
When saving as Pickle, the results are saved in a single Pickle file. The file contains the following dictionary:
event_location_parameters
individual_values
average_values
average_event_properties
average_event_fit
metadata
events
See also
For more information, see the miniml.py save_to_pickle() function.