Aggregate

Summary

The Aggregate class describes an aggregation feature that is applied to a Batch or Stream Feature View via features param.

Example

from tecton import Aggregate, batch_feature_view, TimeWindow
from tecton.types import Int64
from datetime import timedelta


@batch_feature_view(
    # ...
    features=[
        Aggregate(
            column="my_column",
            column_dtype=Int64,
            function="mean",
            time_window=TimeWindow(window_size=timedelta(days=7)),
        ),
        Aggregate(
            column="another_column",
            column_dtype=Int64,
            function="mean",
            time_window=TimeWindow(window_size=timedelta(days=1)),
            name="1d_average",
        ),
    ],
)
def my_fv(data_source):
    pass

Methods

__init__(...)

Parameters

• column (str) – The name of the input/feature column

• column_dtype (SdkDataType) - The data type of the column

• function (Union[str, AggregationFunction]) – One of the built-in aggregation functions. See the aggregation functions reference for a list of aggregation functions.

• time_window (TimeWindow) – The window_size and optional offset over which to aggregate over. See Time Window Reference for more details on the TimeWindow class.

• name (Optional[str]) – The name of this feature, e.g. transaction_count_7d_1d. Defaults to an autogenerated name.