Subtracting one row of data from another in SQL

Doc Falken picture Doc Falken · May 28, 2009 · Viewed 58.3k times · Source

I've been stumped with some SQL where I've got several rows of data, and I want to subtract a row from the previous row and have it repeat all the way down.

So here is the table:

CREATE TABLE foo (
  id,
  length
)
INSERT INTO foo (id,length) VALUES(1,1090)
INSERT INTO foo (id,length) VALUES(2,888)
INSERT INTO foo (id,length) VALUES(3,545)
INSERT INTO foo (id,length) VALUES(4,434)
INSERT INTO foo (id,length) VALUES(5,45)

I want the results to show a third column called difference which is one row subtracting from the one below with the final row subtracting from zero.

+------+------------------------+
| id   |length |  difference  |
+------+------------------------+
|    1 | 1090  |  202         |
|    2 |  888  |  343         |
|    3 |  545  |  111         |
|    4 |  434  |  389         |
|    5 |   45  |   45         |

I've tried a self join but I'm not exactly sure how to limit the results instead of having it cycle through itself. I can't depend that the id value will be sequential for a given result set so I'm not using that value. I could extend the schema to include some kind of sequential value.

This is what I've tried:

SELECT id, f.length, f2.length, (f.length - f2.length) AS difference
FROM foo f, foo f2

Thank you for the assist.

Answer

shahkalpesh picture shahkalpesh · May 28, 2009

This might help you (somewhat).


select a.id, a.length, 
coalesce(a.length - 
    (select b.length from foo b where b.id = a.id + 1), a.length) as diff
from foo a