class svvamp.ExperimentAnalyzer(base_profile=None, n_samples=1, relative_noise=0.0, absolute_noise=0.0, study_profile_criteria=None, voting_rule_tasks=None, output_dir='out', output_file_suffix='', log_csv=None, ping_period=10)[source]#

Analyze a voting experiment.

The results are stored in several csv files.

Parameters:

  • base_profile (Profile, optional) – Profile of the experiment, without added noise.

  • n_samples (int, optional) – Number of samples for the Monte-Carlo treatment.

  • relative_noise (float, optional) – Relative noise used by GeneratorProfileNoise.

  • absolute_noise (float, optional) – Absolute noise used by GeneratorProfileNoise.

  • study_profile_criteria (StudyProfileCriteria, optional) – Profile criteria to be studied.

  • voting_rule_tasks (VotingRuleTasks, optional) – Voting rule tasks to be performed.

  • output_dir (str, optional) – Directory where the outputs will be recorded.

  • output_file_suffix (str, optional) – Suffix for the output file names.

  • log_csv (List of str, optional) – Additional information at the end of lines in the csv file.

  • ping_period (int, optional) – Frequency at which the program prints the number of profiles already analyzed.

Examples

Define an ExperimentAnalyzer with the default options:

>>> experiment_analyzer = ExperimentAnalyzer()

If you specify the profile criteria and/or the voting rule tasks, then the relevant sanity checks are performed:

>>> experiment_analyzer = ExperimentAnalyzer(
...     study_profile_criteria=StudyProfileCriteria(boolean_criteria=['exists_condorcet_winner_rk']),
...     voting_rule_tasks=VotingRuleTasks(voting_systems=[RuleIRV])
... )
StudyProfileCriteria: Sanity check was successful.
VotingRuleTasks: Sanity check was successful.

Anyway, you can then call the object with experiment_analyzer(). You can override the parameters (like base_profile, n_samples, etc) when making the call. For example: experiment_analyzer(base_profile=another_base_profile).