Question

Let’s just say I have two columns, DATE and VALUE. The dates are not necessarily regularly spaced intervals. How can I create a moving average on VALUE over the past N days? I’m using postgres but mysql solutions would also be helpful.

Data:

DATE    VALUE
2012-11-05 10
2012-10-29 31
2012-10-22 108
2012-10-17 3654
2012-10-16 1187
2012-10-15 12033
2012-10-09 41
2012-10-01 85
2012-09-25 20
2012-09-24 285
2012-09-17 20
2012-09-10 20
2012-09-04 41
2012-08-27 63
2012-08-20 52
2012-08-13 160

Answer 1

The MySQL example below covers a sliding 7-day window:

select t1.`DATE`, AVG(t2.`VALUE`) as MV_AVG
from MyTable t1
left outer join MyTable t2 
    on t2.`DATE` between DATE_ADD(t1.`DATE`, INTERVAL -6 DAY) 
        and t1.`DATE`
group by t1.`DATE`

SQL Fiddle Example

Output:

|                             DATE |    MV_AVG |
------------------------------------------------
|    August, 12 2012 20:00:00+0000 |       160 |
|    August, 19 2012 20:00:00+0000 |        52 |
|    August, 26 2012 20:00:00+0000 |        63 |
| September, 03 2012 20:00:00+0000 |        41 |
| September, 09 2012 20:00:00+0000 |      30.5 |
| September, 16 2012 20:00:00+0000 |        20 |
| September, 23 2012 20:00:00+0000 |       285 |
| September, 24 2012 20:00:00+0000 |     152.5 |
| September, 30 2012 20:00:00+0000 |      52.5 |
|   October, 08 2012 20:00:00+0000 |        41 |
|   October, 14 2012 20:00:00+0000 |      6037 |
|   October, 15 2012 20:00:00+0000 |      6610 |
|   October, 16 2012 20:00:00+0000 | 5624.6667 |
|   October, 21 2012 20:00:00+0000 | 1649.6667 |
|   October, 28 2012 20:00:00+0000 |        31 |
|  November, 04 2012 19:00:00+0000 |        10 |