Question

I have a table like this:

---------------------------------------------
|Id | Step                  | StartedAt     |
---------------------------------------------
| 1 | Download Data         | 10:20:00      |
| 2 | Data Quality Control  | 10:45:00      |
| 3 | Run Prediction        | 10:47:00      |
---------------------------------------------

What’s a SQL query that tells me time per each step like this: “Download Data” took 25 minutes, “Data Quality Control” took 2 minutes etc.

Cheers.

P.S My RDBMS is MySQL. Is there any way to do this in MySQL ?

Answer 1

You should use datetime or timestamp instead of time to make this work across date boundaries. See comments.

Standard SQL for most RDBMS

Use a window function for this. Implemented in most notable RDBMS (except MySQL) by now:

SELECT *
     ,lead("StartedAt") OVER (ORDER BY "StartedAt") - "StartedAt" AS duration
FROM  tbl;

lead() retrieves the value of the next row according to the order in the ORDER BY clause. For the last row, where you have no "next" row, you get NULL.

I quote the manual of PostgreSQL on window functions, since you did not name your RDBMS.

MySQL

In the absence of window functions, one way would be to use correlated subqueries:

SELECT t1.*
     ,(SELECT t2."StartedAt"
       FROM tbl t2
       WHERE t2.id > t1.id
       ORDER BY t2."StartedAt"
       LIMIT 1) - "StartedAt" AS duration
FROM  tbl t1;

Or this is probably faster:

SELECT t1."Id", t1."Step", t1."StartedAt"
      ,TIMESTAMPDIFF(MINUTE, t1."StartedAt", min(t2."StartedAt")) AS minutes
FROM   tbl t1
LEFT   JOIN tbl t2 ON t2."Id" > t1."Id"
GROUP  BY t1."Id", t1."Step", t1."StartedAt";

->sqlfiddle with both queries.

The manual about TIMESTAMPDIFF() or TIMEPDIFF()

If your Id column would be ascending without gaps, this would be simpler. But that’s rarely the case in real life tables.