SQL Cartesian Joins

Introduction

In the realm of SQL, the Cartesian join, also known as CROSS JOIN, is a powerful tool for generating paired combinations of rows from two tables. In this blog post, we’ll explore how to leverage Cartesian joins to extract valuable insights from a hypothetical employee_work_location table.

Scenario

Consider a scenario where we have an employee_work_location table with columns such as employee_name, work_location, start_date, and end_date. The goal is to extract data showing employee names and work locations for each month.

Example Data

employee_name	work_location	start_date	end_date
emp_a	Office A	2022-01-01	2022-12-31
emp_b	Office B	2021-05-15	2022-02-28
emp_c	Office C	2022-03-10	2023-01-15
emp_d	Office A	2021-08-20	2022-11-30
emp_a	Office B	2021-03-05	2022-09-15
emp_b	Office C	2022-06-12	2023-03-01
emp_c	Office A	2021-12-01	2022-06-30
emp_d	Office B	2022-02-15	2022-12-15
emp_e	Office C	2021-07-10	2022-08-25
emp_f	Office A	2022-04-18	2023-02-28

Querying with Cartesian Joins

To achieve our goal, we can use a single SQL query with the help of Cartesian joins and a temporary date series. The query generates a series of months from the minimum start date to the maximum end date, performing a cross join with the employee_work_location table. This ensures that we capture all relevant combinations.

SQL Query:

WITH date_series AS (
  SELECT generate_series(
           (SELECT DATE_TRUNC('month', MIN(start_date)) FROM employee_work_location),
           (SELECT DATE_TRUNC('month', MAX(end_date)) + INTERVAL '1 month' FROM employee_work_location),
           '1 month'::interval
         )::DATE AS work_month
)
SELECT
  ds.work_month,
  ew.employee_name,
  ew.work_location
FROM
  date_series ds
CROSS join  employee_work_location ew
where DATE_TRUNC('month', ew.start_date) and DATE_TRUNC('month', ew.end_date)
ORDER BY
  ds.work_month,  ew.employee_name,   ew.work_location;

Note:

We are truncating the month to get the interval from beginning of the month, and adding 1 month to the interval through + INTERVAL '1 month' to not miss the max end month

Result

Now output of the our query presents a comprehensive view of employee work locations for each month.

work_month	employee_name	work_location
2021-03-01	emp_a	Office B
2021-04-01	emp_a	Office B
2021-01-01	emp_a	Office B
2021-06-01	emp_a	Office B
2021-06-01	emp_b	Office B
…	…	…
2022-12-01	emp_b	Office C
2022-12-01	emp_c	Office C
2022-12-01	emp_d	Office B
2022-12-01	emp_f	Office A
2023-01-01	emp_b	Office C
2023-01-01	emp_c	Office C
2023-01-01	emp_f	Office A
2023-02-01	emp_b	Office C
2023-02-01	emp_f	Office A
2023-03-01	emp_b	Office C

Further Enhancements: Separating Months and Years

For additional clarity, we can further refine the result by extracting the year and month using PostgreSQL’s EXTRACT() function.

WITH date_series AS (
  SELECT generate_series(
           (SELECT DATE_TRUNC('month', MIN(start_date)) FROM employee_work_location),
           (SELECT DATE_TRUNC('month', MAX(end_date)) + INTERVAL '1 month' FROM employee_work_location),
           '1 month'::interval
         )::DATE AS work_month
)
SELECT
  extract(year FROM  ds.work_month) as "year",
  extract(month FROM  ds.work_month) as "month",
  ew.employee_name,
  ew.work_location
FROM
  date_series ds
CROSS join  employee_work_location ew
where DATE_TRUNC('month', ds.work_month) between DATE_TRUNC('month', ew.start_date) and DATE_TRUNC('month', ew.end_date)
ORDER BY
  "year", "month",  ew.employee_name,   ew.work_location;

The result now includes separate columns for the year and month, providing a more organized view of the data.

year	month	employee_name	work_location
2021	3	emp_a	Office B
2021	4	emp_a	Office B
2021	4	emp_a	Office B
2021	5	emp_a	Office B
2021	6	emp_a	Office B
…	…	…	…
2023	1	emp_c	Office C
2023	1	emp_f	Office A
2023	2	emp_b	Office C
2023	2	emp_f	Office A
2023	3	emp_b	Office C

With these SQL techniques, we can efficiently extract and organize data, gaining valuable insights into temporal relationships.