Wikipedia talk:Editing frequency

Configuration

Would it be possible for you to do a quick write-up about how you generated these reports? (Scripts and programs used, where data was obtained, etc.) It's incredibly helpful in case you get hit by a bus or suddenly stomp off from the project. :-) --MZMcBride (talk) 18:26, 30 November 2008 (UTC)

1. Download enwiki stub-meta-history.
2. Extract the relevant bits.
3. Write scripts to count.

Extracting the data takes about 80 GB of hard disk space, so not trivial, but still something many people could accomplish. I used Python to write the scripts because it simplified coding the complex data structures. I did have to break up the runs for anons though because Python balked at being asked to store a >1.5 GB data structure in memory. Dragons flight (talk) 20:38, 30 November 2008 (UTC)

Some graphs and analysis

I've been playing with this data and made a few graphs that I think highlight some of the trends. The first is straightforward:

Accounts per month with at least 20 article edits, January 2001 - September 2008

The next one is kind of convoluted, but I think it shows something interesting...that the typical level of participation for active users has declined since the period of rapid growth in 2005 and 2006, but has remained fairly stable in 2007 and 2008:

Ratios of high edit users to all active users since 2005, relative to September 2008 values (article namespace only)

I'm working on an article for the Signpost to summarize the main points of the editing frequency data: User:Ragesoss/Editing frequency stats. Thoughts and/or help are of course welcome.--ragesoss (talk) 02:27, 7 December 2008 (UTC)