WITH T AS (
SELECT '201610' DT , '10059' ID ,	4 CNT FROM DUAL UNION ALL
SELECT '201611', '10059',	6 FROM DUAL UNION ALL
SELECT '201612', '10059',	3 FROM DUAL UNION ALL
SELECT '201708', '10059',	1 FROM DUAL UNION ALL
SELECT '201610', '1021',	20 FROM DUAL UNION ALL
SELECT '201611', '1021',	23 FROM DUAL UNION ALL
SELECT '201612', '1021',	15 FROM DUAL UNION ALL
SELECT '201610', '10280',	13 FROM DUAL UNION ALL
SELECT '201611', '10280',	8 FROM DUAL UNION ALL
SELECT '201612', '10280',	5 FROM DUAL UNION ALL
SELECT '201610', '10369',	4 FROM DUAL UNION ALL
SELECT '201611', '10369',	4 FROM DUAL UNION ALL
SELECT '201612', '10369',	8 FROM DUAL UNION ALL
SELECT '201708', '10369',	2 FROM DUAL UNION ALL
SELECT '201610', '10460',	15 FROM DUAL UNION ALL
SELECT '201611', '10460',   14 FROM DUAL UNION ALL
SELECT '201612', '10460',	24 FROM DUAL UNION ALL
SELECT '201708', '10460',	7 FROM DUAL UNION ALL
SELECT '201708', '11714',	8 FROM DUAL 
)

SELECT CASE WHEN ID IS NULL THEN '합게'
            WHEN DT IS NULL THEN '소계'
            ELSE DT
       END DT , ID , SUM(CNT) CNT
  FROM T
 GROUP BY  ROLLUP (ID ,DT)
 ORDER BY ID
 

SELECT DECODE(GROUPING_ID(mem_no, w_date), 0, w_date, 1, '소계', 3, '합계') w_date
     , mem_no
     , SUM(cnt) cnt
     , DECODE(GROUPING(w_date), 1, ROUND(AVG(cnt), 2)) avg
  FROM (SELECT TO_CHAR(w_date, 'yyyymm') w_date
             , mem_no
--             , COUNT(*) cnt
             , COUNT(DISTINCT TO_CHAR(w_date, 'yyyymmdd')) cnt
          FROM 방문기록테이블
         WHERE TO_CHAR(w_date, 'yyyymm') IN ('201610', '201611', '201612', '201708')
         GROUP BY TO_CHAR(w_date, 'yyyymm'), mem_no
        )
 GROUP BY ROLLUP(mem_no, w_date)
;

네 마농님 말씀하신데로 해당일에 여러번 들어왔어도 한번만 카운트하게 했습니다.

정말 도움 감사합니다~

참고로 유용하게 활용하겠습니다~~^^