V07 Use a pointblank agent

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.

V07a Switch the quality engine

Goal: Run the same formula with pointblank.

Prerequisites: H06; pointblank; extras.

dr_set_engine selects the implementation of a rule, not the storage backend.

training_step(
  "V07a",
  {
    pointblank_rule <- dr_quality_rule("nonnegative", ~ premium >= 0) |>
      dr_set_engine("pointblank")
    pointblank_product <- dr_product(
      "policies.pointblank",
      bad_policies,
      contract = policy_contract
    ) |>
      dr_add_quality(pointblank_rule)
    pointblank_run <- dr_run(
      pointblank_product,
      write = FALSE,
      stop_on_failure = FALSE
    )
    print(dr_quality_report(pointblank_run))
  },
  needs = c("H06"),
  packages = c("pointblank"),
  enabled = training_options$extras,
  optional = TRUE
)

Expected result: The negative premium is still detected.

Task: Has the lake switched to pointblank?

V07 Use a pointblank agent

Goal: Reuse existing pointblank checks.

Prerequisites: V07a.

An agent builder is supplied through a rule adapter. The action remains explicitly blocking.

training_step(
  "V07",
  {
    agent_rule <- dr_pointblank_checks(
      "premium_agent",
      function(data) {
        pointblank::create_agent(data) |>
          pointblank::col_vals_gte(columns = "premium", value = 0)
      },
      action = "block",
      threshold = 0
    )
    agent_product <- dr_product(
      "policies.agent",
      bad_policies,
      contract = policy_contract
    ) |>
      dr_add_quality(agent_rule)
    agent_run <- dr_run(agent_product, write = FALSE, stop_on_failure = FALSE)
    print(dr_quality_report(agent_run))
    agent_contract <- dr_contract(
      "agent.report",
      c(policy_id = "integer", premium = "numeric", cancelled = "logical"),
      rules = list(agent_rule)
    )
    agent_quality <- dr_validate(
      bad_policies,
      agent_contract,
      keep_agents = TRUE
    )
    dr_pointblank_report(
      agent_quality,
      "premium_agent",
      file.path(training_root, "pointblank.html"),
      overwrite = TRUE
    )
  },
  needs = c("V07a"),
  enabled = training_options$extras,
  optional = TRUE
)

Expected result: A pointblank report about the invalid premium.

Task: Must all native rules be replaced?