PartitioningConfig

class datarobotx.PartitioningConfig(User=None, Group=None, DateTime=None, validation_type=None, cv_method=None, validation_pct=None, holdout_pct=None, disable_holdout=None, reps=None)

Partitioning configuration

Parameters that default to ‘None’ (or are omitted by the user) are overridden to server-side defaults at runtime. Consult the DataRobot REST API and GUI documentation for additional information on each parameter.

Parameters:

• User (dict or PartitioningUserConfig) – User partitioning configuration

• Group (dict or PartitioningGroupConfig) – Group partitioning configuration

• DateTime (dict or PartitioningDateTimeConfig) – Date-time partitioning configuration

• validation_type ({'CV', 'TVH'}) – The validation method to be used. CV for cross validation or TVH for train-validation-holdout split.

• cv_method ({'random', 'user', 'stratified', 'group', 'datetime'}) – The partitioning method to be applied to the training data.

• validation_pct (float) – The percentage of the dataset to assign to the validation set

• holdout_pct (float) – The percentage of the dataset to assign to the holdout set

• disable_holdout (bool) – Whether to suppress allocating a holdout fold. If disableHoldout is set to true, holdoutStartDate and holdoutDuration must not be set.

• reps (int) – The number of cross validation folds to use.

See also

DRConfig

Configuration object for DataRobot project and autopilot settings, also includes detailed examples of usage

Attributes:

cv_method	The partitioning method to be applied to the training data.
DateTime	Date-time partitioning configuration
disable_holdout	Whether to suppress allocating a holdout fold.
Group	Group partitioning configuration
holdout_pct	The percentage of the dataset to assign to the holdout set
reps	The number of cross validation folds to use.
User	User partitioning configuration
validation_pct	The percentage of the dataset to assign to the validation set
validation_type	The validation method to be used.

Inherited methods:

keys()
to_dict()	Return configuration as a dict

property DateTime: PartitioningDateTimeConfig

Date-time partitioning configuration

Notes

DateTime : dict or PartitioningDateTimeConfig

Return type:

PartitioningDateTimeConfig

property Group: PartitioningGroupConfig

Group partitioning configuration

Notes

Group : dict or PartitioningGroupConfig

Return type:

PartitioningGroupConfig

property User: PartitioningUserConfig

User partitioning configuration

Notes

User : dict or PartitioningUserConfig

Return type:

PartitioningUserConfig

property cv_method: str

The partitioning method to be applied to the training data.

Notes

cv_method : {‘random’, ‘user’, ‘stratified’, ‘group’, ‘datetime’}

Return type:

str

property disable_holdout: bool

Whether to suppress allocating a holdout fold. If disableHoldout is set to true, holdoutStartDate and holdoutDuration must not be set.

Notes

disable_holdout : bool

Return type:

bool

property holdout_pct: float

The percentage of the dataset to assign to the holdout set

Notes

holdout_pct : float

Return type:

float

property reps: int

The number of cross validation folds to use.

Notes

reps : int

Return type:

int

to_dict()

Return configuration as a dict

property validation_pct: float

The percentage of the dataset to assign to the validation set

Notes

validation_pct : float

Return type:

float

property validation_type: str

The validation method to be used. CV for cross validation or TVH for train-validation-holdout split.

Notes

validation_type : {‘CV’, ‘TVH’}

Return type:

str