Sourced primarily from real UKG time-clock data (falls back to shift-report inference only for an employee with no cached time-clock data). One line per absence type actually seen in the last 6 months, plus Present % and Unexplained %. Updates with the employee-type filter above.