Question

I have a table, Measures, with columns: timestamp (Unix timestamp), tag, value.
To get a single moving average of say, 15 days, I can do something like:

SELECT tag,avg(value) 
FROM measures 
 WHERE tag='xtr' 
 AND timestamp<1353304800
 AND timestamp>1350622800 
 GROUP BY tag;

Now I want to get the moving average of the 150 latest rows, but I’m not sure how to query it.

SELECT t, s,avg(bv) 
FROM 
  (SELECT A.timestamp as t,A.tag as ta,A.value as ac, B.timestamp as tb,B.value as bv, 
  FROM measures 
  CROSS JOIN measures B 
  WHERE A.tag='xtr' AND B.tag='xtr' 
  GROUP BY A.tag) WHERE ROWNUM <= 150;

This is obviously wrong, but I’ve been thinking about it and cant figure it out. Any ideas?
My train of thought is that I need to match each entry with the 150 entries below it and compute the value average for those 150 entries. I’m also pretty sure that there’s probably a better way without a CROSS JOIN, as that would be very slow.

Answer 1

Not entirely clear what your table structure and uniqueness of data is, but this query will give you a true moving average of the previous 150 rows (inclusive of current row) for all timestamps.

SELECT
  tag,
  timestamp,
  avg(value) over (partition by tag
                   order by timestamp asc
                   rows between 149 preceding and current row) moving_avg
FROM
  measures 
WHERE
  tag='xtr' 

If you just need to isolate the most recent 150 rows then base your query on:

select
  tag,
  value
from(
  select tag
         value
  from measures
  order by timestamp desc)
where
  rownum <= 150