Distinct count in a column based on another column in sql - sql

Question

I have an employee table with two columns: emp_id and month_of_project. I want to find out the distinct count of employees involved in a project for a particular month. Meaning if the same person is involved in 3 months we will only count that person for the first month. I have mentioned sample below
emp_id month_of_project
101 Jan
102 Jan
103 Jan
101 Feb
104 Mar
102 Mar
105 Apr
103 Apr
The result should be
month count
Jan 3
Feb 0
Mar 1
Apr 1
Is there any way to achieve this in sql?

Answer 1

I think that you only should use GROUP BY clause
SELECT month_of_project, COUNT(emp_id)
FROM employees
GROUP BY month_of_project

Answer 2

You could use NOT EXISTS to only fetch records where no record in a previous month exists. Too bad you chose a textual representation for the month, not a numerical one. So you first have to translate it into numbers. You can use a CASE expression here.
SELECT t1.month_of_project,
count(*)
FROM elbat t1
WHERE NOT EXISTS (SELECT *
FROM elbat t2
WHERE CASE t2.month_of_project
WHEN 'Jan' THEN
1
...
WHEN 'Dec' THEN
12
END
<
CASE t1.month_of_project
WHEN 'Jan' THEN
1
...
WHEN 'Dec' THEN
12
END
AND t2.emp_id = t1.emp_id)
GROUP BY t1.month_of_project;

Answer 3

The sql is for oracle. And please don't store month like this in your real system. Use a date field. If your use case is just month then year can be arbitrary like 2000 and day can be 01 but atleaset for sorting etc. having true date always always always is the right idea.
First with is just to simulate data. Second in the with part mons is to get list of possible months since you want 0 for Feb. If you have that table outside you don't need that.
Then basically inner query finds the first month for employee as the month to use and outer query counts distinct
with emp_mon as
(
select '101' emp_id, to_date('20190101','YYYYMMDD') month_of_project from dual
union all select '102', to_date('20190101','YYYYMMDD') from dual
union all select '103', to_date('20190101','YYYYMMDD') from dual
union all select '101', to_date('20190201','YYYYMMDD') from dual
union all select '104', to_date('20190301','YYYYMMDD') from dual
union all select '102', to_date('20190301','YYYYMMDD') from dual
union all select '105', to_date('20190401','YYYYMMDD') from dual
union all select '103', to_date('20190401','YYYYMMDD') from dual
),
mons as
(
select distinct month_of_project
from emp_mon
)
select mons.month_of_project, count(distinct emp_first_mon.emp_id) cnt_emp_id
from mons
left outer join
(
select emp_id, min(month_of_project) month_of_project
from emp_mon
group by emp_id
) emp_first_mon on emp_first_mon.month_of_project = mons.month_of_project
group by mons.month_of_project
order by 1

Answer 4

SQL Sever COUNT DISTINCT will do exactly what you want. It's important to group on the correct column however.
WITH TEMP AS
(
SELECT 1 AS EMP, 1 AS MONTH_D
UNION ALL
SELECT 2 AS EMP, 1 AS MONTH_D
UNION ALL
SELECT 2 AS EMP, 1 AS MONTH_D
UNION ALL
SELECT 3 AS EMP, 1 AS MONTH_D
UNION ALL
SELECT 1 AS EMP, 2 AS MONTH_D
)
SELECT MONTH_D, COUNT(DISTINCT EMP) FROM TEMP
GROUP BY MONTH_D