Oracle SQL Pivot Command Not Working -- "Invalid Idenitifier" - sql

Question

I am trying to do a simple query in Oracle SQL using Pivot. I have 12 columns in my table, but only the first four are of interest. The columns are "FACILITY_ID" (Unique Identifier), "REPORTING_YEAR" (Year data was reported), "SUBPART" (Category of data wanting to pivot by - values acceptable between A and Z), and "CO2E" (floating point number reported). I want to take only the data from subparts C and K (the third column) and add them up into their own columns.
For example:
If my original table looks like:
FACILITY_ID --- REPORTING_YEAR --- SUBPART --- CO2E
10 -------------- 2020 ---------------------- C -------- 10
11 -------------- 2020 ---------------------- K -------- 20
10 -------------- 2020 ---------------------- K -------- 40
10 -------------- 2020 ---------------------- K -------- 40
11 -------------- 2020 ---------------------- C -------- 30
I would like to get something like:
FACILITY_ID --- REPORTING_YEAR ----- C ----- K
10 -------------- 2020 ----------------- 10 ----- 80
11 -------------- 2020 ----------------- 30 ----- 20
My code as of now is as follows:
Create Table CandK_emissions as
Select
FACILITY_ID,
REPORTING_YEAR,
SUBPART,
CO2E
From
facilityReport
Pivot
(
Sum(CO2E)
For SUBPART
in ('C',
'K')
)
The error I'm getting is 'Ora-00904: "CO2e": Invalid Identifier'. I've double checked that everything is spelled correctly. I'm not sure where I'm going wrong.

Answer 1

The SELECT clause must show what columns will exist in the output, not what columns are used from the base table. In your query, you are aggregating the CO2E values, separately for different values of SUBPART. The result of pivoting doesn't have CO2E values anymore, or SUBPART values; instead, the two SUBPART values you are interested in, C and K, become two distinct columns in the output, and the sums of CO2E appear in these two new columns.
Your query should SELECT columns C and K; the error message is telling you that after pivoting, there is no CO2E column to select. (After you change that, it would tell you the same thing about SUBPART.)
Here is how this should look. Note a few things: In the "test data" I simulated one more column, a primary key named PK, to demonstrate that you must EXCLUDE it first, in an inline view (the sub-SELECT in the FROM clause). Also note that, to make the PIVOT as efficient as possible, in the subquery I only select the rows we need - those where the SUBPART is either C or K. We don't need to see the other rows.
The WITH clause is just for testing - remove it, and use your actual table and column names in the query.
with
test_data (pk, facility_id, reporting_year, subpart, co2e) as (
select 335, 10, 2020, 'C', 10 from dual union all
select 440, 11, 2020, 'K', 20 from dual union all
select 482, 10, 2020, 'K', 40 from dual union all
select 106, 10, 2020, 'K', 40 from dual union all
select 476, 11, 2020, 'C', 30 from dual
)
-- end of test data; actual query begins below this line
select facility_id, reporting_year, c, k
from (
select facility_id, reporting_year, subpart, co2e
from test_data
where subpart in ('C', 'K')
)
pivot (sum(co2e) for subpart in ('C' as c, 'K' as k))
order by facility_id, reporting_year -- or whatever is needed
;
FACILITY_ID REPORTING_YEAR C K
----------- -------------- ---------- ----------
10 2020 10 80
11 2020 30 20
Note - it may be simpler (and more efficient) to do away with PIVOT altogether, and to "pivot" the old way, using conditional aggregation (the way pivoting was done before the PIVOT operator was introduced). Something like this:
select facility_id, reporting_year,
sum(case subpart when 'C' then co2e end) as c,
sum(case subpart when 'K' then co2e end) as k
from test_data
group by facility_id, reporting_year
order by facility_id, reporting_year
;