nvtabular.ops.Groupby#

class nvtabular.ops.Groupby(groupby_cols=None, sort_cols=None, aggs='list', name_sep='_', ascending=True)[source]#

Bases: Operator

Groupby Transformation

Locally transform each partition of a Dataset with one or more groupby aggregations.

WARNING: This transformation does NOT move data between partitions. Please make sure that the target Dataset object is already shuffled by groupby_cols, otherwise the output may be incorrect. See: Dataset.shuffle_by_keys.

Example usage:

groupby_cols = ['user_id', 'session_id']
dataset = dataset.shuffle_by_keys(keys=groupby_cols)

groupby_features = [
    'user_id', 'session_id', 'month', 'prod_id',
] >> ops.Groupby(
    groupby_cols=groupby_cols,
    sort_cols=['month'],
    aggs={
        'prod_id': 'list',
        'month': ['first', 'last'],
    },
)
processor = nvtabular.Workflow(groupby_features)

workflow.fit(dataset)
dataset_transformed = workflow.transform(dataset)
Parameters:
  • groupby_cols (str or list of str) – The column names to be used as groupby keys. WARNING: Ensure the dataset was partitioned by those groupby keys (see above for an example).

  • sort_cols (str or list of str) – Columns to be used to sort each partition before groupby aggregation is performed. If this argument is not specified, the results will not be sorted.

  • aggs (dict, list or str) – Groupby aggregations to perform. Supported list-based aggregations include “list”, “first” & “last”. Most conventional aggregations supported by Pandas/cuDF are also allowed (e.g. “sum”, “count”, “max”, “mean”, etc.).

  • name_sep (str) – String separator to use for new column names.

__init__(groupby_cols=None, sort_cols=None, aggs='list', name_sep='_', ascending=True)[source]#

Methods

__init__([groupby_cols, sort_cols, aggs, ...])

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[, ...])

Provides a hook method for sub-classes to override to implement custom column selection logic.

create_node(selector)

export(path, input_schema, output_schema, ...)

Export the class object as a config and all related files to the user defined path.

inference_initialize(col_selector, model_config)

Configures this operator for use in inference.

load_artifacts([artifact_path])

Load artifacts from disk required for operator function.

output_column_names(col_selector)

Given a set of columns names returns the names of the transformed columns this operator will produce

save_artifacts([artifact_path])

Save artifacts required to be reload operator state from disk

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

dynamic_dtypes

export_name

Provides a clear common english identifier for this operator.

is_subgraph

label

output_dtype

output_properties

output_tags

supported_formats

supports

Returns what kind of data representation this operator supports

transform(col_selector: ColumnSelector, df: DataFrame) DataFrame[source]#

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_output_schema(input_schema: Schema, col_selector: ColumnSelector, prev_output_schema: Optional[Schema] = None) Schema[source]#
column_mapping(col_selector)[source]#
property dependencies#