nvtabular.ops.JoinExternal

class nvtabular.ops.JoinExternal(df_ext, on, how='left', on_ext=None, columns_ext=None, drop_duplicates_ext=None, kind_ext=None, cache='host', **kwargs)

Bases: nvtabular.ops.operator.Operator

Join each dataset partition to an external table. For performance reasons, only “left” and “inner” join transformations are supported.

Example usage:

# Load dataset which should be joined to the main dataset
df_external = cudf.read_parquet('external.parquet')

# Use JoinExternal to define a NVTabular workflow
joined = ColumnSelector(columns_left) >> nvt.ops.JoinExternal(
    df_ext,
    on=['key1', 'key2'],
    on_ext=['key1_ext', 'key2_ext'],
    how='left',
    columns_ext=['key1_ext', 'key2_ext', 'cat1', 'cat2', 'num1'],
    kind_ext='cudf',
    cache='device'
) >> ...
processor = nvtabular.Workflow(joined)

Parameters

• df_ext (DataFrame, pyarrow.Table, Dataset, dd.DataFrame, or file path(s)) – The external table to join to each partition of the dataset. Note that the join must be a partition-wise transformation. Therefore, if df_ext is a multi-partition Dask collection, it will need to be broadcasted to every partition.

• on (str or list(str)) – Column name(s) to merge on

• how ({"left", "inner"}; default "left") – Type of join operation to perform.

• on_ext (str or list(str); Optional) – Column name(s) on external table to join on. By default, we assume on_ext is the same as on.

• columns_ext (list(str); Optional) – Subset of columns to select from external table before join.

• drop_duplicates_ext (bool; Default False) – Drop duplicates from external table before join.

• kind_ext (ExtData; Optional) – Format of df_ext. If nothing is specified, the format will be inferred.

• cache ({"device", "host", "disk"}) – Where to cache df_ext between transformations. Only used if the data is originally stored on disk. The “host” option is also supported when df_ext is a cudf.DataFrame.

__init__(df_ext, on, how='left', on_ext=None, columns_ext=None, drop_duplicates_ext=None, kind_ext=None, cache='host', **kwargs)

Methods

__init__(df_ext, on[, how, on_ext, …])
column_mapping(col_selector)
compute_column_schema(col_name, input_schema)
compute_input_schema(root_schema, …)	Given the schemas coming from upstream sources and a column selector for the input columns, returns a set of schemas for the input columns this operator will use
compute_output_schema(input_schema, col_selector)
compute_selector(input_schema, selector, …)
create_node(selector)
inference_initialize(col_selector, model_config)	Configures this operator for use in inference.
output_column_names(col_selector)	Given a set of columns names returns the names of the transformed columns this operator will produce
transform(col_selector, df)	Transform the dataframe by applying this operator to the set of input columns
validate_schemas(parents_schema, …[, …])	Provides a hook method that sub-classes can override to implement schema validation logic.

Attributes

dependencies	Defines an optional list of column dependencies for this operator.
dynamic_dtypes
label
output_dtype
output_properties
output_tags
supports	Returns what kind of data representation this operator supports

transform(col_selector: merlin.dag.selector.ColumnSelector, df: pandas.core.frame.DataFrame) → pandas.core.frame.DataFrame

Transform the dataframe by applying this operator to the set of input columns

Parameters

• columns (list of str or list of list of str) – The columns to apply this operator to

• df (Dataframe) – A pandas or cudf dataframe that this operator will work on

Returns

Returns a transformed dataframe for this operator

Return type

DataFrame

compute_selector(input_schema: merlin.schema.schema.Schema, selector: merlin.dag.selector.ColumnSelector, parents_selector: merlin.dag.selector.ColumnSelector, dependencies_selector: merlin.dag.selector.ColumnSelector) → merlin.dag.selector.ColumnSelector

compute_output_schema(input_schema, col_selector, prev_output_schema=None)

column_mapping(col_selector)