https://github.com/wikimedia/analytics-reportcard-data/edit/master/datafiles/rc_page_requests.csv

Data for wikidata in cassandra it's not fetch-able:

cassandra@cqlsh> select * from "local_group_default_T_lgc_pagecounts_per_project".data where "_domain"='analytics.wikimedia.org' and "access-site" in ('desktop-site','mobile-site','all-sites') and granularity in ('hourly', 'daily', 'monthly') and project='www.wikidata' and timestamp ='2016010800';

Returns records:
cassandra@cqlsh> select * from "local_group_default_T_lgc_pagecounts_per_project".data where "_domain"='analytics.wikimedia.org' and "access-site" in ('desktop-site','mobile-site','all-sites') and granularity in ('hourly', 'daily', 'monthly') and project='www.wikidata' and timestamp ='2016010800';

_domain | project | access-site | granularity | timestamp | _tid | _del | count
-------------------------+--------------+--------------+-------------+------------+--------------------------------------+------+---------
analytics.wikimedia.org | www.wikidata | all-sites | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 4943451
analytics.wikimedia.org | www.wikidata | all-sites | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 188718
analytics.wikimedia.org | www.wikidata | desktop-site | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 4803906
analytics.wikimedia.org | www.wikidata | desktop-site | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 182718
analytics.wikimedia.org | www.wikidata | mobile-site | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 139545
analytics.wikimedia.org | www.wikidata | mobile-site | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 6000

So something is going on with aqs parsing besides the fact that wikidata is persisted like www.wikidata, should be "wikidata"

cassandra@cqlsh> select * from "local_group_default_T_pageviews_per_project".data where "_domain"='analytics.wikimedia.org' and "access" in ('desktop','mobile','all') and granularity in ('hourly', 'daily', 'monthly') and project='wikidata' and timestamp ='2016010800' and agent='user';

_domain | project | access | agent | granularity | timestamp | _tid | _del | v | views
-------------------------+----------+---------+-------+-------------+------------+--------------------------------------+------+--------+-------
analytics.wikimedia.org | wikidata | desktop | user | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 283772 | null
analytics.wikimedia.org | wikidata | desktop | user | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 6947 | null

Pageviews on meta definitely not real, see wikistats: (that interval is removed) https://stats.wikimedia.org/wikispecial/EN/ReportCardTopWikis.htm#lang_meta

mmm.. is this right?

hive (wmf)> select * from domain_abbrev_map where hostname like '%meta%';
OK
domain_abbrev_map.domain_abbrev domain_abbrev_map.hostname domain_abbrev_map.access_site
meta.m meta.wikimedia.org desktop
meta.m.m meta.wikimedia.org mobile
meta.zero.m meta.wikimedia.org zero

Also wikidata seems to have issues but the map seems ok:

www.wd www.wikidata.org desktop
m.wd www.wikidata.org mobile
zero.wd www.wikidata.org zero

Data from projectcounst_raw, note the meta.m and meta.mw sites have spikes but the one present on reportcard lines up with data on meta.m, that site, according to our abbreviation scheme is the meta desktop domain.

Spike is "real" on table so i guess we need to dig files to see where the misstranslation is happening.

At the time of loading:
www.wikidata.org needs to be changed to wikidata
www.mediawiki.org needs to be changed to mediawiki

And code needs to be corrected ( i think) on javascript so those two are parsed correctly

Change 346802 had a related patch set uploaded (by Nuria):
[analytics/refinery@master] Corrected triming of hostname

https://gerrit.wikimedia.org/r/346802

Realoaded data (took about 2 hours) and now wikidata project column looks correct on db:

cassandra@cqlsh> select * from "local_group_default_T_lgc_pagecounts_per_project".data where "_domain"='analytics.wikimedia.org' and "access-site" in ('desktop-site','mobile-site','all-sites') and granularity in ('hourly', 'daily', 'monthly') and project='wikidata' and timestamp ='2016010800';

_domain | project | access-site | granularity | timestamp | _tid | _del | count
-------------------------+----------+--------------+-------------+------------+--------------------------------------+------+---------
analytics.wikimedia.org | wikidata | all-sites | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 4943451
analytics.wikimedia.org | wikidata | all-sites | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 188718
analytics.wikimedia.org | wikidata | desktop-site | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 4803906
analytics.wikimedia.org | wikidata | desktop-site | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 182718
analytics.wikimedia.org | wikidata | mobile-site | daily | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 139545
analytics.wikimedia.org | wikidata | mobile-site | hourly | 2016010800 | 13814000-1dd2-11b2-8080-808080808080 | null | 6000

Change 346802 merged by Nuria:
[analytics/refinery@master] Correcting loading of pagecounts into cassandra

https://gerrit.wikimedia.org/r/346802

I found this in DammitSummarizeProjectviews.pl:

1. quick fix: fake counts for meta for a period where we had > 10 billion hits on meta due to fundraiser artefact, all to wiki/Special:RecordImpression

if (($language =~ /^meta/) && (($yyyymm ge '201208') && ($yyyymm le '201504')))
{

next if $language =~ /^meta\./ ; # ignore mobile and zero, set fixed count for desktop
$count = sprintf ("%.0f",7000000 / (24 * 30)) ;

}