功能工具 - RollingSum

Question

我正在尝试使用功能工具创建自定义原语 rolling-sum 功能，下面是代码：-

class RollingSumOnDatetime(TransformPrimitive):
    """Calculates the rolling sum on a Datetime time index column.
    Description:
        Given a list of values and a Datetime time index, return the rolling sum.
    """

    name = "rolling_sum_on_datetime"
    input_types = [Numeric, DatetimeTimeIndex]
    return_type = Numeric
    uses_full_entity = True
    description_template = "the rolling sum of {} on {}"

    def __init__(self, window=None,on=None):
        self.window = window
        self.on = on


    def get_function(self):
        def rolling_sum(to_roll, on_column):
            """method is passed a pandas series"""
            # create a DataFrame that has the both columns in it
            df = pd.DataFrame({to_roll.name: to_roll, on_column.name: on_column})
            rolled_df = df.rolling(window=self.window, on=on_column.name).sum()
            return rolled_df[to_roll.name]

        return rolling_sum


feature_matrix, feature_defs = ft.dfs(
            entityset=es,
            n_jobs=10,
            target_entity="contracts",
            agg_primitives=agg_prim,
            trans_primitives=trans_prim,
            groupby_trans_primitives=[
                RollingSumOnDatetime(window="5D", on=es["days"]["datetime"])
            ],
            max_depth=2,
            drop_contains=["contract_id", "merchant_id"],
        )

代码的第一部分是自定义原语，第二部分我调用函数 它给出了错误：

ValueError: setting an array element with a sequence.

Answer 1

当您将原语传递给groupby_trans_primitives 时，您需要删除on=es["days"]["datetime"]。它不是RollingSumOnDatetime 的__init__ 中的参数，因此不适用。

这是一个最小的、可重现的例子：

from featuretools.primitives import AggregationPrimitive, TransformPrimitive
from featuretools.variable_types import Numeric, DatetimeTimeIndex

class RollingSumOnDatetime(TransformPrimitive):
    """Calculates the rolling sum on a Datetime time index column.
    Description:
        Given a list of values and a Datetime time index, return the rolling sum.
    """
    name = "rolling_sum_on_datetime"
    input_types = [Numeric, DatetimeTimeIndex]
    return_type = Numeric
    uses_full_entity = True
    description_template = "the rolling sum of {} on {}"
    def __init__(self, window=None):
        self.window = window

    def get_function(self):
        def rolling_sum(to_roll, on_column):
            """method is passed a pandas series"""
            #create a DataFrame that has the both columns in it
            df = pd.DataFrame({to_roll.name:to_roll, on_column.name:on_column})
            rolled_df = df.rolling(window=self.window, on=on_column.name).sum()
            return rolled_df[to_roll.name]
        return rolling_sum

import featuretools as ft 

es = ft.demo.load_mock_customer(return_entityset=True)

feature_matrix, feature_defs = ft.dfs(
    entityset=es,
    target_entity="transactions",
    agg_primitives=[],
    trans_primitives=[],
    groupby_trans_primitives=[
        RollingSumOnDatetime(window="5D")
    ]
)
feature_defs

如果我们打印出feature_defs，我们会得到：

[<Feature: session_id>,
 <Feature: amount>,
 <Feature: product_id>,
 <Feature: ROLLING_SUM_ON_DATETIME(amount, transaction_time, window=5D) by product_id>,
 <Feature: ROLLING_SUM_ON_DATETIME(amount, transaction_time, window=5D) by session_id>,
 <Feature: products.brand>,
 <Feature: sessions.customer_id>,
 <Feature: sessions.device>,
 <Feature: sessions.customers.zip_code>,
 <Feature: ROLLING_SUM_ON_DATETIME(amount, sessions.session_start, window=5D) by product_id>,
 <Feature: ROLLING_SUM_ON_DATETIME(amount, sessions.session_start, window=5D) by session_id>,
 <Feature: ROLLING_SUM_ON_DATETIME(amount, sessions.session_start, window=5D) by sessions.customer_id>]