H24 Group by broker

Use your current course session. Preparation and course runner · R script ↓

H24 Group by broker

Goal: Reuse a metric with a different grouping.

Prerequisites: H23.

dimensions permits a dimension; by selects it for this measurement.

training_step(
  "H24",
  {
    broker_rates <- dataraft.metrics::dr_measure(
      cohort_run,
      cancellation_metric,
      by = "broker_id"
    )
    print(broker_rates)
    stopifnot(nrow(broker_rates) == 2L)
    stopifnot(isTRUE(all.equal(sort(broker_rates$value), c(0, 1 / 3))))
  },
  needs = c("H23")
)

Expected result: Broker 1 has a rate of 1/3; broker 2 has 0.

Task: Can subgroup rates always be averaged directly?