seldonian.spec.HyperparameterSelectionSpec

class HyperparameterSelectionSpec(hyper_schema, n_bootstrap_trials, n_bootstrap_workers, use_bs_pools, confidence_interval_type=None)

Bases: object

Class for the specification object for selecting hyperparameters

Parameters:

  • hyper_schema (seldonian.hyperparam_search.HyperSchema) – A hyperparameter schema specifying which hyperparameters to tune and the values to sweep over.

  • n_bootstrap_trials (int) – The number of bootstrap trials to run

  • n_bootstrap_workers (int) – The number of workers (parallel processes) to use when running bootstrap trials

  • use_bs_pools (bool) – Whether to use sampling pools during bootstrapping

  • confidence_interval_type (str) – “ttest” or “clopper-pearson”

__init__(hyper_schema, n_bootstrap_trials, n_bootstrap_workers, use_bs_pools, confidence_interval_type=None)
__repr__()

Return repr(self).

Methods