DeepchecksModelVersionClient.get_production_data#
- DeepchecksModelVersionClient.get_production_data(start_time: Union[datetime, str, int], end_time: Union[datetime, str, int], rows_count: int = 10000, filters: List[DataFilter] = None, deepchecks_format: bool = False) Union[DataFrame, Tuple[Dataset, Optional[ndarray], Optional[ndarray]]]#
Get DataFrame or Deepchecks dataset and predictions for a model version production data on a specific window.
- Parameters
- model_version_idint
The model version id.
- start_timet.Union[datetime, str, int]
- The start time timestamp.
int: Unix timestamp
str: timestamp in ISO8601 format
datetime: If no timezone info is provided on the datetime assumes local timezone.
- end_timet.Union[datetime, str, int]
- The end time timestamp.
int: Unix timestamp
str: timestamp in ISO8601 format
datetime: If no timezone info is provided on the datetime assumes local timezone.
- rows_countint, optional
The number of rows to return (random sampling will be used).
- filterst.List[DataFilter], optional
Data filters to apply. Used in order to received a segment of the data based on selected properties. Required format for filters and possible operators are detailed under the respected objects which can be found at: from deepchecks_client import DataFilter, OperatorsEnum
- deepchecks_formatbool, default False
If True will return in Deepchecks format: (Deepchecks dataset, predictions array, prediction probabilities array)
- Returns
- t.Union[‘pandas’.DataFrame, t.Tuple[Dataset, t.Optional[np.ndarray], t.Optional[np.ndarray]]]
The production dataframe or if deepchecks_format is True - a tuple of: (Deepchecks dataset, predictions array, prediction probabilities array).