python - 将参数传递给 featuretools 中的聚合基元

Question

我正在使用 featuretools dfs 函数使用与时间相关的聚合原语(例如 TimeSince ( https://docs.featuretools.com/api_reference.html#deep-feature-synthesis ))生成我的 feature_matrix/p>

feature_matrix, feature_defs = ft.dfs(entityset=es,
                                     target_entity="users",
                                     max_depth=2,
                                     agg_primitives=["sum", "max", "min", 
                                                     "mean", "median", "count", 
                                                     "avg_time_between"],
                                     trans_primitives=["day", "year", "month", "weekday", 
                                                       "time_since_previous", "time_since",
                                                      ],
                                     cutoff_time=cutoff_times,
                                     cutoff_time_in_index=True)

如何将 [unit] 参数传递给 trans_primitives 参数以将时间单位更改为分钟、小时等？

Answer 1

要将参数传递给聚合或转换原语，您可以使用参数创建原语的实例。这是将 unit 参数传递给 TimeSince 原语的方法。

from featuretools.primitives import TimeSince

ft.dfs(
    ...
    trans_primitives=[TimeSince(unit='hours')]
)