FSD distribution of German page lengths of a sample (with length > 1000)

SQL

use dewiki_p;
select year, fsd, count(*) as frequency
from 
(select missing_year as year, page_id, substr(max(rev_len),1,1) as fsd
from
revision,
(select page_id, year(max(rev_timestamp)) as last_year
from revision, page
where rev_page = page_id
and page_namespace = 0
and page_id % 100 = 0
and rev_len > 1000
group by page_id) last_revision,
(select 2018 as missing_year union select 2017 union select 2016 union select 2015 union select 2014 union select 2013 union select 2012 union select 2011 union select 2010 union select 2009 union select 2008 union select 2007 union select 2006 union select 2005 union select 2004 union select 2003 union select 2002 ) all_years
where missing_year > last_year
and rev_page = page_id
and year(rev_timestamp) = last_year
group by missing_year, page_id
union
select year(a.rev_timestamp) as year, page_id, substr(max(a.rev_len), 1, 1) as fsd
from revision a, revision b, revision c, revision d, page
where 
a.rev_page = page_id
and page_id % 100 = 0
and a.rev_len > 1000
 and a.rev_parent_id = b.rev_id
and b.rev_parent_id = c.rev_id
and c.rev_parent_id = d.rev_id
and abs(cast(a.rev_len as double) - cast(d.rev_len as double)) / cast(d.rev_len as double) < 0.2
and page_namespace = 0
group by year(a.rev_timestamp), page_id) distribution
group by year, fsd;
SQL

Checking query status...
