fitbenchmarking.results_processing.performance_profiler module¶
Set up performance profiles for both accuracy and runtime tables
-
fitbenchmarking.results_processing.performance_profiler.
create_plot
(ax, step_values, solvers)¶ Function to draw the profile on a matplotlib axis
Parameters: - ax (an .axes.SubplotBase subclass of ~.axes.Axes (or a subclass of ~.axes.Axes)) – A matplotlib axis to be filled
- step_values (list of float) – a sorted list of the values of the metric being profiled
- solvers (list of strings) – A list of the labels for the different solvers
-
fitbenchmarking.results_processing.performance_profiler.
plot
(acc, runtime, fig_dir)¶ Function that generates profiler plots
Parameters: - acc (dict) – acc dictionary containing number of occurrences
- runtime (dict) – runtime dictionary containing number of occurrences
- fig_dir (str) – path to directory containing the figures
Returns: path to acc and runtime profile graphs
Return type: tuple(str, str)
-
fitbenchmarking.results_processing.performance_profiler.
prepare_profile_data
(results)¶ Helper function which generates acc and runtime dictionaries which contain number of occurrences that the minimizer produces a normalised result which is less than the bounds in PROFILER_BOUNDS
Parameters: results (list of list of fitbenchmarking.utils.fitbm_result.FittingResult) – results nested array of objects Returns: dictionary containing number of occurrences Return type: tuple(dict, dict)
-
fitbenchmarking.results_processing.performance_profiler.
profile
(results, fig_dir)¶ Function that generates profiler plots
Parameters: - results (list of list of fitbenchmarking.utils.fitbm_result.FittingResult) – results nested array of objects
- fig_dir (str) – path to directory containing the figures
Returns: path to acc and runtime profile graphs
Return type: tuple(str, str)