Using PL/SQL Analytics to Count on Substring?

Question

How should I write query to return a count of 3 for Farms Beginning with K ?

Why does (partition by id,substr(farm,1)) calculate to 1

with tree_harvest
as ( 
select 1 as id, 'PINE' as tree, 'K001' as farm from dual union all
select 1 as id, 'PINE' as tree, '0003' as farm from dual union all
select 1 as id, 'PINE' as tree, 'K002' as farm from dual union all
select 1 as id, 'PINE' as tree, 'K003' as farm from dual
)
select id, tree,farm,
       count(*) over (partition by id) as id_count,
       case
       when regexp_like(farm,'^K','i') 
       then count(*) over (partition by id,substr(farm,1))
       else 0
       end as k_count
       from tree_harvest;

   
   

Desired Result

 ID TREE FARM   ID_COUNT  K_COUNT  
  1 PINE  0003  4         0
  1 PINE  K001  4         3
  1 PINE  K002  4         3
  1 PINE  K003  4         3

Answer 1

Here is a solution that fixes your issue, and should be faster (more efficient) than your current approach. Note that here both analytic functions partition by id only; the conditional count is taken care separately, within the count() call itself. Also both comparisons to K or k are case insensitive; in your attempted query one of the comparisons wasn't. I also avoid regular expressions (slower), which are not needed here.

with tree_harvest
as ( 
select 1 as id, 'PINE' as tree, 'K001' as farm from dual union all
select 1 as id, 'PINE' as tree, '0003' as farm from dual union all
select 1 as id, 'PINE' as tree, 'K002' as farm from dual union all
select 1 as id, 'PINE' as tree, 'K003' as farm from dual
)
select id, tree,farm,
       count(*) over (partition by id) as id_count,
       case when lower(farm) like 'k%' then
           count(case when lower(farm) like 'k%' then 1 end) 
                over (partition by id) else 0 end as k_count
       from tree_harvest;
       
        ID TREE FARM   ID_COUNT    K_COUNT
---------- ---- ---- ---------- ----------
         1 PINE K001          4          3
         1 PINE K003          4          3
         1 PINE K002          4          3
         1 PINE 0003          4          0