culebra.tools module¶
Tools to automate the execution of experiments.
This module is composed by:
The
abcsub-module, where some abstract base classes are defined to support the evaluation of trainers and decision managers that let select an adequate solutionThe
evaluationsub-module, providing several classes to make automated experimentation easier.The
decision_managersub-module, which offers several decision managers.The
Datasetclass to hold and manage the datasets.The
EffectSizeclass, to keep the outcome of an effect size estimation of several batches resultsThe
Resultsclass, to manage the results provided by the evaluation of anyTrainerThe
ResultsAnalyzerclass, to perform statistical analysis over the results of several experimtent batchsThe
ResultsComparisonclass, to keep the outcome of a comparison of several batches resultsThe
TestOutcomeclass, to keep the outcome of a statistical test
Attributes¶
- DEFAULT_ALPHA = 0.05¶
Default significance level for statistical tests.
- DEFAULT_CONFIG_SCRIPT_FILENAME = 'config.py'¶
Default file name for configuration files.
- DEFAULT_EXCEL_FILE_EXTENSION = '.xlsx'¶
File extension for Excel datasheets.
- DEFAULT_HOMOSCEDASTICITY_TEST = <function bartlett>¶
Default homoscedasticity test.
- DEFAULT_OUTLIER_PROPORTION = 0.05¶
Expected outlier proportion por class.
- DEFAULT_NORMALITY_TEST = <function shapiro>¶
Default normality test.
- DEFAULT_NUM_EXPERIMENTS = 1¶
Default number of experiments in the batch.
- DEFAULT_P_ADJUST = 'fdr_tsbky'¶
Default method for adjusting the p-values with the Dunn’s test.
- DEFAULT_RESULTS_BASE_FILENAME = 'results'¶
Default base name for results files.
- DEFAULT_RUN_SCRIPT_FILENAME = 'run.py'¶
Default file name for the script to run an evaluation.
- DEFAULT_SCRIPT_FILE_EXTENSION = '.py'¶
Default file extension for python scripts.
- DEFAULT_SEP = '\\s+'¶
Default column separator used within dataset files.

