Recipes
- class datarobot.models.recipe.Recipe(dialect, recipe_id, status, inputs, operations=None, downsampling=None, settings=None)
Data wrangling entity, which contains all information needed to transform dataset and generate SQL.
- classmethod update_downsampling(recipe_id, downsampling)
Set downsampling for the recipe, applied during publishing.
- Return type:
- retrieve_preview(max_wait=600, number_of_operations_to_use=None)
Retrieve preview and compute it, if absent.
- Parameters:
- max_wait: int
The number of seconds to wait for the result.
- number_of_operations_to_use: int, optional
Request preview for particular number of operations.
- Returns:
- preview: dict
- Return type:
Dict[str,Any]
- retrieve_insights(max_wait=600, number_of_operations_to_use=None)
Retrieve insights for the sample. When preview is requested, the insights job starts automatically.
- Parameters:
- max_wait: int
The number of seconds to wait for the result.
- number_of_operations_to_use: int, optional
Retrieves insights for the specified number of operations. First, preview computation for the same number of operations must be submitted.
- Returns:
- Return type:
Any
- classmethod set_operations(recipe_id, operations)
Set operations for the recipe.
- Return type:
- get_sql(operations=None)
Generate sql for the given recipe in a transient way, recipe is not modified. if operations is None, recipe operations are used to generate sql. if operations = [], recipe operations are ignored during sql generation. if operations is not empty list, generate sql for them.
- Return type:
str
- classmethod from_data_store(use_case, data_store, data_source_type, dialect, data_source_inputs)
Create a wrangling recipe from data store.
- Return type:
- class datarobot.models.recipe.RecipeSettings(target=None, weights_feature=None)
Settings, for example to apply at downsampling stage.
- class datarobot.models.recipe.RecipeDatasetInput(input_type, dataset_id, dataset_version_id, sampling=None, alias=None)
Object, describing inputs for recipe transformations.
- class datarobot.models.recipe.DatasetInput(sampling)
- class datarobot.models.recipe.DataSourceInput(canonical_name, table, schema=None, catalog=None, sampling=None)
Inputs required to create a new recipe from data store.
Recipe Operations
- class datarobot.models.recipe_operation.WranglingOperation(directive, arguments)
- class datarobot.models.recipe_operation.DownsamplingOperation(directive, arguments)
- class datarobot.models.recipe_operation.SamplingOperation(directive, arguments)
- class datarobot.models.recipe_operation.BaseTimeAwareTask(name, arguments)
- class datarobot.models.recipe_operation.TaskPlanElement(column, task_list)
- class datarobot.models.recipe_operation.CategoricalStats(methods, window_size)
- class datarobot.models.recipe_operation.NumericStats(methods, window_size)
- class datarobot.models.recipe_operation.Lags(orders)
- class datarobot.models.recipe_operation.LagsOperation(column, orders, datetime_partition_column, multiseries_id_column=None)
Generate lags in a window.
- class datarobot.models.recipe_operation.WindowCategoricalStatsOperation(column, window_size, methods, datetime_partition_column, multiseries_id_column=None, rolling_most_frequent_udf=None)
Generate rolling statistics in a window for categorical features.
- class datarobot.models.recipe_operation.WindowNumericStatsOperation(column, window_size, methods, datetime_partition_column, multiseries_id_column=None, rolling_median_udf=None)
Generate various rolling numeric statistics in a window. Output could be a several columns.
- class datarobot.models.recipe_operation.TimeSeriesOperation(target_column, datetime_partition_column, forecast_distances, task_plan, baseline_periods=None, known_in_advance_columns=None, multiseries_id_column=None, rolling_median_udf=None, rolling_most_frequent_udf=None, forecast_point=None)
Operation to generate a dataset ready for time series modeling: with forecast point, forecast distances, known in advance columns, etc.
- class datarobot.models.recipe_operation.ComputeNewOperation(expression, new_feature_name)
- class datarobot.models.recipe_operation.RenameColumnsOperation(column_mappings)
- class datarobot.models.recipe_operation.FilterCondition
- class datarobot.models.recipe_operation.FilterOperation(conditions, keep_rows=True, operator='and')
Filter rows.
- class datarobot.models.recipe_operation.DropColumnsOperation(columns)
- class datarobot.models.recipe_operation.RandomSamplingOperation(rows, seed=None)
- class datarobot.models.recipe_operation.DatetimeSamplingOperation(datetime_partition_column, rows, strategy=None, multiseries_id_column=None, selected_series=None)