Creating counts based on date ranges with inner join - sql

Question

Here is an illustration for what I'd like to do
Table A:
user_id | industry | startdate | enddate | generation
1 retail 2000-01-01 2001-01-01 Gen X
1 retail 2002-01-01 2003-02-01 Gen X
2 Tech 2001-01-01 2002-01-01 Gen X
2 Business 2002-03-01 2003-01-01 Gen X
2 Tech 2003-02-01 null Gen X
... ... ... ... ...
35642 Medicine 2020-02-01 2022-03-01 Gen Z
Table B
month
1990-01-01
1990-02-01
...
2022-03-01
Desired Result:
industry | generation| count | month
retail Gen X 200 2002-02-01
retail Gen Y 250 2002-02-01
Tech Gen X 130 2002-02-01
Tech Gen Y 166 2002-02-01
...
For now, I've only got tables A and B. I want to create counts by industry, by month, by generation, but I'm not sure how I can do this using the two tables that I have.
My (incorrect) approach would be something like select count(*), industry, month, generation where A.startdate < B.month and A.enddate > B.month, but this query is obviously not running. Is what I want to do possible with just tables A and B?
Apologies if I'm being unclear, I am admittedly new to SQL queries and am not sure how to approach this problem.

Answer 1

Try this approach:
Create a CTE that generates the distinct list of industry/generation by querying table A
Create a 2nd CTE that cartesian joins the first CTE to table B - giving you a list of all months/industry/generation
Left outer join table A to the 2nd CTE and query for the result you want to achieve

Answer 2

Executing the steps of NickW gives me.
I guess that users whose enddate is null are still in businss
## Create tables
CREATE TABLE table_a (
user_id int,
industry text,
startdate date,
enddate date,
generation text
);
CREATE TABLE table_b (
month date
);
# Insert data
WITH series_months AS (
SELECT date(i)
from generate_series(
date '1999-01-01',
date '2012-09-01',
INTERVAL '1 month'
) i
)
INSERT INTO table_b (month)
SELECT * FROM series_months;
INSERT INTO table_a (user_id, industry, startdate, enddate, generation)
VALUES
(1, 'retail', '2000-01-01', '2001-01-01', 'Gen X'),
(1, 'retail', '2002-01-01', '2003-02-01', 'Gen X'),
(2, 'Tech', '2001-01-01', '2002-01-01', 'Gen X'),
(2, 'Business', '2002-03-01', '2003-01-01', 'Gen X'),
(2, 'Tech', '2003-02-01', NULL, 'Gen X');
# Perform joins
WITH industry_generation as (
SELECT distinct industry, generation from table_a
), months_industry_generation as (
SELECT *
FROM table_b, industry_generation
), combined_table AS (
SELECT mig.month, mig.industry, mig.generation, user_id, startdate, enddate
FROM
(SELECT * FROM months_industry_generation) mig
inner join table_a a
on mig.industry = a.industry AND mig.generation = a.generation
where month >= startdate AND (month <= enddate OR enddate is null)
)
SELECT industry, generation, count(user_id) AS count, month
FROM combined_table
GROUP BY month, industry, generation
ORDER BY 1, 2, 4
;