Row_number calculation in sql, oracle - sql

Question

Now i am facing one small issue in sql code.
I have no idea about this logic.
with res as (
select
85,
2,
45,
34,
60,
2,
6,
4,
78,
6,
23,
45,
80,
80
from dual
)
select * from res
I have mention 3 header .1.input 2.Expected 3.Actual.
And my question is set one column based on input days we have to display rn column. IF value is days > 60 them we have to display as zero and next row value will be start as 1 and to printed as 2, 3, 4... till days > 60.then we have to set as 0 for days > 60 and agin start from 1, 2,3 ... still days > 60. its loop till end of value.
Input Expected Actual
DAYS DAYS RN DAYS RN
85 85 0 85 1
4 4 1 4 2
32 32 2 32 3
7 7 3 7 4
5 5 4 5 5
66 66 0 66 6
14 14 1 14 7
25 25 2 25 8
2 2 3 2 9
9 9 4 9 10
70 70 0 70 11
80 80 0 80 12
6 6 1 6 13
3 3 2 3 14
1 1 3 1 15
78 78 0 78 16

Answer 1

OK, first of all... for meaningful results, you need a way to order the input data to get a row number for each record. You cannot just rely on each record's position in a list, because that's undefined in Oracle if these records are coming from a table.
So, assuming we add a rn column to your res table, all you need to do is this:
select res.rn,
res.days,
res.rn - nvl(last_value(case
when res.days < 60 then null
else res.rn end) ignore nulls over (order by rn),0) lrn
from res;
That nvl(last_value... expression finds the rn (row number) of the most recent input record having days >= 60. So, if a record's rn is 15 and the most recent record with days >= 60 was 13, then that record gets renumbered as "2".
For this to work, your rn values in the input data must start at 1 and have no gaps. If that is not the case with your real data, then you will need to add another with clause before this to do a dense_rank or something to get them that way.
Here is a complete example, using your test data (from your expected result):
with res (rn, days) as (
select 1, 85 from dual union all
select 2, 4 from dual union all
select 3, 32 from dual union all
select 4, 7 from dual union all
select 5, 5 from dual union all
select 6, 66 from dual union all
select 7, 14 from dual union all
select 8, 25 from dual union all
select 9, 2 from dual union all
select 10, 9 from dual union all
select 11, 70 from dual union all
select 12, 80 from dual union all
select 13, 6 from dual union all
select 14, 3 from dual union all
select 15, 1 from dual union all
select 16, 78 from dual
)
-- Above is just test data.. solution starts here
/* with ... */
select res.rn, res.days, res.rn - nvl(last_value(
case when res.days < 60 then null else res.rn end)
ignore nulls over (order by rn),0) lrn
from res
+----+------+-----+
| RN | DAYS | LRN |
+----+------+-----+
| 1 | 85 | 0 |
| 2 | 4 | 1 |
| 3 | 32 | 2 |
| 4 | 7 | 3 |
| 5 | 5 | 4 |
| 6 | 66 | 0 |
| 7 | 14 | 1 |
| 8 | 25 | 2 |
| 9 | 2 | 3 |
| 10 | 9 | 4 |
| 11 | 70 | 0 |
| 12 | 80 | 0 |
| 13 | 6 | 1 |
| 14 | 3 | 2 |
| 15 | 1 | 3 |
| 16 | 78 | 0 |
+----+------+-----+
LRN is the "expected" RN from your desired results. For simplicity, I omitted the "actual" columns, which just seemed to be duplicates of the input data.

Answer 2

You don't include your actual select statement.
I'm guessing it is something like this:
select res.*, row_number() OVER () AS RN from res
You want to use modulus like this:
select res.*, (row_number() OVER ()) % 5 AS RN from res