with  t1
as (
select '1001' idx,'1' user_id,'50000' amount,to_date('2016-11-12','yyyy-mm-dd') as dt from dual union all
select '2001','2','30000',to_date('2016-12-13','yyyy-mm-dd') from dual union all
select '3001','2','20000',to_date('2017-01-13','yyyy-mm-dd') from dual union all
select '4001','1','40000',to_date('2017-02-13','yyyy-mm-dd') from dual union all
select '5001','3','40000',to_date('2017-03-16','yyyy-mm-dd') from dual union all
select '6001','4','30000',to_date('2017-04-16','yyyy-mm-dd') from dual union all
select '7001','3','20000',to_date('2017-05-17','yyyy-mm-dd') from dual 
)
,t2
as (
select '1' user_id,'홍길동' user_name from dual union all
select '2','김나나' from dual union all
select '3','송하나' from dual union all
select '4','미니언즈' from dual
)
select a.idx, b.user_name, a.amount, a.dt from
(
select idx, user_id, amount, dt, row_number() over(partition by user_id order by dt, idx) rn 
from t1) a
,t2 b
where a.user_id = b.user_id  
and dt between to_date('2017-01-01','yyyy-mm-dd') and to_date('2017-03-31','yyyy-mm-dd')
and rn = 1

위의 방법으로 하면 테이블이 작으면 괜찮을 것 같은데 테이블 사이즈가 아주 크면

전체 기간에 대해 user_id별로 순번을 구해야해서 비효율이 있겠네요.

select idx, user_name, amount, dt
  from
    (
    select a.idx, b.user_name, a.amount, a.dt, row_number() over(partition by a.user_id order by dt, idx) rn 
      from t1 a, t2 b
     where a.user_id = b.user_id
       and dt between to_date('2017-01-01','yyyy-mm-dd') and to_date('2017-03-31','yyyy-mm-dd')
       and not exists (select 1 
                         from t1 b 
                        where b.user_id = a.user_id
                          and dt < to_date('2017-01-01','yyyy-mm-dd'))
    ) 
 where rn = 1

이렇게 하면 좀 나을 것 같습니다.