class datarobotx.Configurator(base)#

Synthesize and apply common DR configuration patterns.

As with drx.DRConfig, parameters that are omitted or None will be assigned default values automatically by DataRobot as required.

Methods apply configuration to base and then return a new instance of Configurator to allow chaining.


base (AutopilotModel or DRConfig) – Base model or configuration object to apply configuration to.


otv([n_backtests, validation_duration, holdout])

DataRobot OTV configuration.

otv(n_backtests=None, validation_duration=None, holdout=None)#

DataRobot OTV configuration.

When using OTV the datetime_partition_column keyword argument indicating the column name of the primary datetime feature must be subsequently passed to fit().

  • n_backtests (int) – The number of backtests folds to use

  • validation_duration (str or datetime.timedelta) – The duration of the validation dataset within each backtest fold. String will be parsed by pandas.Timedelta()

  • holdout (bool or tuple of (str or datetime.datetime, str or datetime.timedelta)) – If False, the holdout fold will be disabled. Otherwise, must be a tuple of strings denoting (holdout_start, holdout_duration). Start date will be parsed by pandas.to_datetime() and duration by pandas.Timedelta() if specified as strings.


New Configurator object with the OTV configuration applied

Return type: