V24 Review in the Data Viewer

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

Optional deep dive. Enable only the options named in each block, after preparing its packages or services.

V24a Inspect column lineage

Goal: Read static output-column dependencies.

Prerequisites: H08; extras.

Static analysis can be incomplete for dynamic R code. It is not presented as complete knowledge.

training_step(
  "V24a",
  {
    column_recipe <- dr_recipe() |>
      dr_step_mutate(annual_premium = premium * 12) |>
      dr_step_select(policy_id, annual_premium)
    print(dr_column_lineage(column_recipe, names(policies)))
  },
  needs = c("H08"),
  enabled = training_options$extras,
  optional = TRUE
)

Expected result: annual_premium depends on premium.

Task: Must the recipe execute for this analysis?

V24b Select diagnostics

Goal: Separate run status from rule execution errors.

Prerequisites: H06; extras.

These functions inspect retained results without starting another run.

training_step(
  "V24b",
  {
    print(dr_status(failed_run))
    print(dr_quality_errors(failed_run))
    print(dr_quality_counts(n_failed = 1, n_total = 6))
  },
  needs = c("H06"),
  enabled = training_options$extras,
  optional = TRUE
)

Expected result: Blocked status. quality_errors may be empty because FALSE is not an R exception. quality_counts constructs counts for custom rules; it does not count a run automatically.

Task: Is a failure rate of 1/6 automatically acceptable?

V24c Use quality as a test expectation

Goal: Reuse the quality decision in tests.

Prerequisites: H05; testthat; extras.

dr_expect_quality uses the framework's quality decision in automated tests.

training_step(
  "V24c",
  {
    quality_test_run <- dr_run(policy_product, write = FALSE)
    dr_expect_quality(quality_test_run)
  },
  needs = c("H05"),
  packages = c("testthat"),
  enabled = training_options$extras,
  optional = TRUE
)

Expected result: No test failure for valid data.

Task: Should an intentional test failure go uncaught in the main path?

V24 Review in the Data Viewer

Goal: Open diagnostics interactively.

Prerequisites: H06; gui=TRUE; interactive R session.

The viewer shows bounded failing rows. No UI starts in batch mode.

training_step(
  "V24",
  {
    if (interactive()) {
      dr_review(failed_run, what = "rows", limit = 10)
    }
    if (interactive()) dr_review(failed_run, what = "report")
  },
  needs = c("H06"),
  enabled = training_options$gui,
  optional = TRUE
)

Expected result: Failing rows or a quality report in the viewer.

Task: Does this start a new run?