SQL: How to use DECODE and COUNT functions together - sql

Question

I'm working on a homework problem, but we haven't learned about the DECODE function, only CASE. This week's unit is about using aggregate functions. Below is my homework question and what my professor wants as result-
"The following two questions are very challenging (you need to use DECODE function to complete them).
Create a query that will display the total number of employees and of that total
the number who were hired in the year 1980, 1981, 1982, and 1987. Give
appropriate column headings. (5 Points)
Total 1980 1981 1982 1987
----- ----- ----- ----- -----
14 1 10 1 2
Here is the function I typed in the server along with result. I only tried the year 1980 so I don't waste time, but I also need 1981, 1982, and 1987-
SELECT COUNT(ename) AS "Total",
COUNT(DECODE(hiredate, '80', '1980'))
FROM emp;
Total COUNT(DECODE(HIREDATE,'80','1980'))
----- -----------------------------------
14 0
Here is the datatype for the 'hiredate' column-
HIREDATE
---------
17-NOV-81
01-MAY-81
09-JUN-81
02-APR-81
28-SEP-81
20-FEB-81
08-SEP-81
03-DEC-81
22-FEB-81
03-DEC-81
17-DEC-80
HIREDATE
---------
09-DEC-82
12-JAN-83
23-JAN-82
14 rows selected.
Thank you for anyone's help!

Answer 1

Assuming that the hiredate column has the data type DATE.
Lets do it without DECODE and use PIVOT instead:
Oracle Setup:
CREATE TABLE emp ( ename, hiredate ) AS
SELECT 'A', DATE '1980-01-01' FROM DUAL UNION ALL
SELECT 'B', DATE '1981-01-01' FROM DUAL UNION ALL
SELECT 'C', DATE '1981-02-01' FROM DUAL UNION ALL
SELECT 'D', DATE '1981-03-01' FROM DUAL UNION ALL
SELECT 'E', DATE '1981-04-01' FROM DUAL UNION ALL
SELECT 'F', DATE '1981-05-01' FROM DUAL UNION ALL
SELECT 'G', DATE '1981-06-01' FROM DUAL UNION ALL
SELECT 'H', DATE '1981-07-01' FROM DUAL UNION ALL
SELECT 'I', DATE '1981-08-01' FROM DUAL UNION ALL
SELECT 'J', DATE '1981-09-01' FROM DUAL UNION ALL
SELECT 'K', DATE '1981-10-01' FROM DUAL UNION ALL
SELECT 'L', DATE '1982-01-01' FROM DUAL UNION ALL
SELECT 'M', DATE '1987-01-01' FROM DUAL UNION ALL
SELECT 'N', DATE '1987-02-01' FROM DUAL;
Query:
SELECT *
FROM (
SELECT EXTRACT( YEAR FROM hiredate ) AS hireyear,
COUNT(*) OVER () AS "Total"
FROM emp
)
PIVOT ( COUNT(*) FOR hireyear IN ( 1980, 1981, 1982, 1987 ) )
Output:
Total | 1980 | 1981 | 1982 | 1987
----: | ---: | ---: | ---: | ---:
14 | 1 | 10 | 1 | 2
Query 2:
You can also do it with CASE:
SELECT COUNT(*) AS "Total",
COUNT(
CASE
WHEN hiredate >= DATE '1980-01-01' AND hiredate < DATE '1981-01-01'
THEN hiredate
END
) AS "1980",
COUNT(
CASE
WHEN hiredate >= DATE '1981-01-01' AND hiredate < DATE '1982-01-01'
THEN hiredate
END
) AS "1981",
COUNT(
CASE
WHEN hiredate >= DATE '1982-01-01' AND hiredate < DATE '1983-01-01'
THEN hiredate
END
) AS "1982",
COUNT(
CASE
WHEN hiredate >= DATE '1987-01-01' AND hiredate < DATE '1988-01-01'
THEN hiredate
END
) AS "1987"
FROM emp
Query 3:
Or with DECODE.
SELECT COUNT(*) AS "Total",
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1980, 'anything here' ) ) AS "1980",
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1981, 'anything here' ) ) AS "1981",
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1982, 'anything here' ) ) AS "1982",
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1987, 'anything here' ) ) AS "1987"
FROM emp
Note: When you are counting, you count that the value passed to COUNT is not-NULL so the DECODE function can return any non-NULL value when it matches and it will get counted; conversely, so long as the value returned to COUNT is NULL, which is the default for DECODE when you don't supply an extra even-numbered argument, then it won't count the row.
So you could use any literal value. Like a string:
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1980, 'anything here' ) ) AS "1980"
or a number
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1980, 1 ) ) AS "1980"
or even the hiredate column (which if it has a year of 1980 then you know it isn't NULL):
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1980, hiredate ) ) AS "1980"
and if you want to be explicit about the return value from DECODE when it doesn't match then put in an extra NULL argument:
COUNT( DECODE( EXTRACT( YEAR FROM hiredate ), 1980, 42, NULL ) ) AS "1980"
db<>fiddle here

Answer 2

Your teacher wants you to do conditional aggregation:
SELECT
COUNT(ename) AS "Total",
SUM(DECODE(hire_date, 1980, 1, 0)) AS "1980",
SUM(DECODE(hire_date, 1981, 1, 0)) AS "1981",
SUM(DECODE(hire_date, 1982, 1, 0)) AS "1982",
SUM(DECODE(hire_date, 1987, 1, 0)) AS "1987"
FROM emp;
Within each SUM, DECODE() checks the hire_date against the target value, and counts in only records that have the relevant date for each column.
This assumes that hire_date is actually a year, which seems counterintuitive. If it is a DATE, then you would need to extrat the year part, like so:
SELECT
COUNT(ename) AS "Total",
SUM(DECODE(EXTRACT(YEAR from hire_date), 1980, 1, 0)) AS "1980",
SUM(DECODE(EXTRACT(YEAR from hire_date), 1981, 1, 0)) AS "1981",
SUM(DECODE(EXTRACT(YEAR from hire_date), 1982, 1, 0)) AS "1982",
SUM(DECODE(EXTRACT(YEAR from hire_date), 1987, 1, 0)) AS "1987"
FROM emp;
Please note that DECODE() is an Oracle-specific function that is not supported by other RDBMS. A more standard way to write this is to use CASE blocks, with expressions like:
SUM(CASE WHEN EXTRACT(YEAR from hire_date) = 1980 THEN 1 ELSE 0 END) as "1980"

Answer 3

select "1981"+"1982"+"1983" Total, "1981","1982","1983" from (select
count(case to_char(hiredate,'YYYY') when '1981' then 1 end) "1981",
count(case to_char(hiredate,'YYYY') when '1982' then 1 end) "1982",
count(case to_char(hiredate,'YYYY') when '1983' then 1 end) "1983"
from emp);