Grupowanie zmiennych binarnych według identyfikatora i dat min./maks.

Od Oracle 12 jest to typ problemu, który MATCH_RECOGNIZE ma na celu rozwiązanie:

SELECT *
FROM   table_name
MATCH_RECOGNIZE (
  PARTITION BY Emp_ID
  ORDER     BY dt
  MEASURES
    FIRST(dt) AS min_date,
    LAST(dt)  AS max_date,
    MATCH_NUMBER() AS group_number
  PATTERN (one+)
  DEFINE
    one AS dummy = 1
)

Co dla przykładowych danych:

CREATE TABLE table_name ( dt, Emp_ID, Dummy ) AS
SELECT DATE '2021-01-01', 5, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-02', 5, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-03', 5, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-04', 5, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-05', 5, 0 FROM DUAL UNION ALL
SELECT DATE '2021-01-06', 5, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-07', 5, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-01', 8, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-02', 8, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-03', 8, 1 FROM DUAL UNION ALL
SELECT DATE '2021-01-04', 8, 0 FROM DUAL UNION ALL
SELECT DATE '2021-01-05', 8, 0 FROM DUAL UNION ALL
SELECT DATE '2021-01-06', 8, 0 FROM DUAL UNION ALL
SELECT DATE '2021-01-07', 8, 1 FROM DUAL;

Wyjścia:

db<>fiddle tutaj