Get Average value for each X rows in SQL

Question

Let`s say I have the following table

+----+-------+
| Id | Value |
+----+-------+
|  1 |   2.0 |
|  2 |   8.0 |
|  3 |   3.0 |
|  4 |   9.0 |
|  5 |   1.0 |
|  6 |   4.0 |
|  7 |   2.5 |
|  8 |   6.5 |
+----+-------+

I want to plot these values, but since my real table has thousands of values, I thought about getting and average for each X rows. Is there any way for me to do so for, ie, each 2 or 4 rows, like below:

2
+-----+------+
| 1-2 |  5.0 |
| 3-4 |  6.0 |
| 5-6 |  2.5 |
| 7-8 |  4.5 |
+-----+------+

4
+-----+------+
| 1-4 |  5.5 |
| 5-8 |  3.5 |
+-----+------+

Also, is there any way to make this X value dynamic, based on the total number of rows in my table? Something like, if I have 1000 rows, the average will be calculated based on each 200 rows (1000/5), but if I have 20, calculate it based on each 4 rows (20/5).

I know how to do that programmatically, but is there any way to do so using pure SQL?

EDIT: I need it to work on mysql.

Answer 1

Depending on your DBMS, something like this will work:

SELECT
   ChunkStart = Min(Id),
   ChunkEnd = Max(Id),
   Value = Avg(Value)
FROM
   (
      SELECT
         Chunk = NTILE(5) OVER (ORDER BY Id),
         *
      FROM
         YourTable
   ) AS T
GROUP BY
   Chunk
ORDER BY 
   ChunkStart;

This creates 5 groups or chunks no matter how many rows there are, as you requested.

If you have no windowing functions you can fake it:

SELECT
   ChunkStart = Min(Id),
   ChunkEnd = Max(Id),
   Value = Avg(Value)
FROM
   YourTable
GROUP BY
   (Id - 1) / (((SELECT Count(*) FROM YourTable) + 4) / 5)
;

I made some assumptions here such as Id beginning with 1 and there being no gaps, and that you would want the last group too small instead of too big if things didn't divide evenly. I also assumed integer division would result as in Ms SQL Server.

Answer 2

You can use modulos operator to act on every Nth row of the table. This example would get the average value for every 10th row:

select avg(Value) from some_table where id % 10 = 0;

You could then do a count of the rows in the table, apply some factor to that, and use that value as a dynamic interval:

select avg(Value) from some_table where id % (select round(count(*)/1000) from some_table) = 0;

You'll need to figure out the best interval based on the actual number of rows you have in the table of course.

EDIT: Rereading you post I realize this is getting an average of every Nth row, and not each sequential N rows. I'm not sure if this would suffice, or if you specifically need sequential averages.