Message boards : Number crunching : Wrong Website Statistics
Message board moderation

To post messages, you must log in.

AuthorMessage
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jun 08
Posts: 1608
Credit: 94,642,557
RAC: 98,659
Message 36286 - Posted: 7 Aug 2018, 4:35:41 UTC

At least some of the project statistic pages show wrong values as there are users with a (much) higher application average than total average.

Examples from
https://lhcathome.cern.ch/lhcathome/per_app_list.php?appid=14&is_team=&is_total=0

User: xxxxxxx
ATLAS (recent average): 78,196.99
total (recent average): 0.10

User: xxxxxxx
ATLAS (recent average): 42,408.63
total (recent average): 1,538.30

User: xxxxxxx
ATLAS (recent average): 24,496.37
total (recent average): 0.07

User: xxxxxxx
ATLAS (recent average): 16,186.81
total (recent average): 0.07
ID: 36286 · Report as offensive     Reply Quote
Harri Liljeroos
Avatar

Send message
Joined: 28 Sep 04
Posts: 484
Credit: 25,910,910
RAC: 14,597
Message 36298 - Posted: 7 Aug 2018, 9:45:58 UTC

My guess is that application averages are calculated with different formula than total recent average. Maybe it is even outside the Boinc server code? There seems to be a much longer decay time on application average than on total recent average.
ID: 36298 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jun 08
Posts: 1608
Credit: 94,642,557
RAC: 98,659
Message 36301 - Posted: 7 Aug 2018, 9:59:06 UTC - in response to Message 36298.  

You may be right, although I don't see any sense behind different decay times.

In addition you may look at CMS.
Lots of users with a recent average of several thousand credits although there was no work for months.
ID: 36301 · Report as offensive     Reply Quote
Harri Liljeroos
Avatar

Send message
Joined: 28 Sep 04
Posts: 484
Credit: 25,910,910
RAC: 14,597
Message 36302 - Posted: 7 Aug 2018, 10:19:40 UTC - in response to Message 36301.  

You may be right, although I don't see any sense behind different decay times.

Nor do I.

In addition you may look at CMS.
Lots of users with a recent average of several thousand credits although there was no work for months.

Yep, I see the same for Theory although I haven't done any for months.
ID: 36302 · Report as offensive     Reply Quote
computezrmle
Volunteer moderator
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 15 Jun 08
Posts: 1608
Credit: 94,642,557
RAC: 98,659
Message 36391 - Posted: 13 Aug 2018, 6:51:05 UTC

It looks like users who don't deliver any more results remain on their last average forever while users who deliver results, even invalids like for CMS, get their average updated.
ID: 36391 · Report as offensive     Reply Quote

Message boards : Number crunching : Wrong Website Statistics


©2021 CERN