This dataset is described on its dumps download page.
This dataset is a compressed format of the best pageview data that the Wikimedia Foundation had at any point in its historyː
- From 2007 to December 2015, it compressed the pagecounts-raw dataset, which is now deprecated (providing pageviews per project from December 2007 on, and pageviews per article from late 2011 on)
- From Dec 2015 to Present day, it compresses the pageviews dataset
More information about each of those datasets can be found on their pages.
- Erik Zachte's 2011 announcement of this dataset (source for the description on the dumps download page, link is broken there): ,
- Phabricator task to make per-article data available in pagecounts-ez back to 2008 (instead of 2011)
- m:Learning patterns/Tips for reading project codes from pageviews data files