Skip to content

Time Related Feature Set

Usage

config.yaml
...
features:
  TimeFeaturesConfig:
    months_for_avg: 3

Features Generated

Feature Description
AVG_DAYS_BETWEEN_ORDERS Provides the average amount of time between all orders in the specified date range, will return NONE if an average can not be calculated.
TIME_SINCE_FIRST_ORDER Provides the number of days that have passed since the outlet placed it's first order in the dataset, measured to the calculation date (Tenure).
TIME_SINCE_LAST_ORDER Provides the number of days that have passed since the last order the outlet placed in the dataset, measured to the calculation date (Recency).

Feature Configuration

TimeFeaturesConfig Class

TimeFeaturesConfig is a configuration class for generating features based on relative dates within a dataset.

Attributes

  • months_for_avg: Number of months to include from the calculation date when calculating averages.

Feature Generation

TimeFeatureSet Class

TimeFeatureSet is a feature set that generates features based on relative dates within a dataset.

Methods

  • calculate(df, dataset_config, feature_config, calculation_date): Calculate the features for the given DataFrame, DatasetConfig, and FeatureConfig given a specific calculation_date.

Feature Functions

  • time_btwn_orders(df, date_col, keys): Calculate the average time between orders by key columns.
  • tenure(df, date_col, keys, calc_date): Calculate the length of time of relationship to the run date by key columns.
  • time_since_last_order(df, date_col, keys, calc_date): Calculate the time since the last order placed to the run date by key columns.