PL-SQL “LAST_VALUE” function does not give the desired result

Question

I have a table with below structure:

create table DOC_AMOUNT_DETAIL
(
  col_id         NUMBER,
  amount         NUMBER,
  operation_date DATE      
)

and some example data for this table is :

col_id    |  amount   |  operation_date 
---------------------------------------
  1       |  5000     |  11/1/2020  
  2       |  1000     |  11/1/2020 
  3       |  3000     |  11/1/2020     
  1       |  1000     |  11/14/2020
  2       |   500     |  11/14/2020

I'm trying to use last_value function to have below output:

col_id    |  amount   
--------------------
  1       |  1000      
  2       |  500    
  3       |  3000

and the problem is that the query does not give me the desired result :

select t.col_id,
       last_value(t.amount) over(partition by t.col_id order by t.operation_date
        ) amount
from doc_amount_detail t
where t.operation_date <= to_date('11/15/2020', 'mm/dd/yyyy');

Instead, it gives me this :

col_id    |  amount   
--------------------
  1       |  5000    
  1       |  1000
  2       |  1000
  2       |  500
  3       |  3000

Everything seems fine with my query so can anybody please explain where the exact problem is? Thanks in advance

Answer 1

Explaination

The ORDER BY clause (in an analytical function), implies a window clause of ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW.

You need ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING

Since this will apply to each row, you'll need a DISTINCT clause.

Example

with data(col_id, amount, operation_date ) as (
    select  1,  5000,  to_date( '11/1/2020', 'mm/dd/yyyy' ) from dual union all
    select  2,  1000,  to_date( '11/1/2020', 'mm/dd/yyyy' ) from dual union all 
    select  3,  3000,  to_date( '11/1/2020', 'mm/dd/yyyy' ) from dual union all
    select  1,  1000,  to_date( '11/14/2020', 'mm/dd/yyyy' ) from dual union all
    select  2,   500,  to_date( '11/14/2020', 'mm/dd/yyyy' ) from dual
)
select distinct col_id
    , last_value( amount ) over (partition by col_id order by operation_date
                                 rows BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) amt
from data;

Results: