Question

I have a table with the following structure – steamid, itemid, eventid, value , all just ints or big ints that log hourly data I insert. So a user most likely has multiple entries with the same steamids, itemids, and even values.

I’m trying to get the 100 top values – but the same item must not repeat.

What I have so far is

SELECT itemid,value,steamid 
FROM $table 
GROUP BY itemid 
ORDER BY value DESC 
LIMIT 0,100

Which gives me this data set (sample only here):

itemid      value   steamid
=================================================
599291414   66397   76561198032389066
779150329   62882   76561198001229760
773965297   51895   76561198014617403
332883551   43201   76561197992659494
333434836   40880   7656119799359013

However, this for some reason ignores the true largest value listed in the table. I’m not sure how else I can format this so that I can ignore duplicate entries with the same itemid and steamid. I don’t think I can group by steamid because then it would ignore other items that could be associated with a steamid.

Here’s the first few if I select without grouping.

 itemid     value   steamid
 =====================================
 1011809265 753665  76561198010314894
 376615188  101684  76561197989760193
 478937438  83448   76561198010314894
 478937438  83448   76561198010314894
 376662587  72693   76561197989760193
 376662587  72693   76561197989760193
 599291414  66454   76561198032389066
 599291414  66454   76561198032389066
 599291414  66454   76561198032389066

Insight appreciated and I’ll gladly answer any questions that may help in figuring this out.

Answer 1

Try

SELECT * FROM (

  SELECT itemid,value,steamid FROM $table GROUP BY itemid ORDER BY value DESC
) as tbl1

GROUP BY itemid, value

LIMIT 0,100

This should unique itemid, but then give you only the single values.