Kiran Jonnalagadda (![[info]](http://p-stat.livejournal.com/img/userinfo.gif){kind=small} jace) wrote,@ 2006-04-19 20:35:00 |
|
Charting traffic in ![[info]](http://p-stat.livejournal.com/img/community.gif)
bangalore
For my upcoming session at Barcamp, I took the last four years of posts and comments stats forbangalore and made a graph. I’ve never made a graph before and took the obvious route of Excel. This is ugly. It hurts my eyes. Can someone make it pretty? mannu? Here’s the excel sheet, raw data (per post, monthly), and source for downloading calendar pages (sh) and parsing into CSV (py). Four posts were backdated (start here); I manually added them to the excel sheet based on the date of the first comment to each. Deleted posts—which were among the most interesting ones—are unfortunately not countable.
The community’s most active period was between May 2004 and May 2005. Comment spikes appear to correlate with post spikes. If both are redrawn to a percentage scale using daily instead of monthly stats, it should be easier to identify the volatile periods. How do I do that in Excel? What charting app should I use?
That huge spike in April 2005 (1209 comments) is clearly the work of our friends atwearesphinx. It should be interesting to relate other spikes with specific events:
bangaloreFor my upcoming session at Barcamp, I took the last four years of posts and comments stats for
bangalore and made a graph. I’ve never made a graph before and took the obvious route of Excel. This is ugly. It hurts my eyes. Can someone make it pretty? mannu? Here’s the excel sheet, raw data (per post, monthly), and source for downloading calendar pages (sh) and parsing into CSV (py). Four posts were backdated (start here); I manually added them to the excel sheet based on the date of the first comment to each. Deleted posts—which were among the most interesting ones—are unfortunately not countable.The community’s most active period was between May 2004 and May 2005. Comment spikes appear to correlate with post spikes. If both are redrawn to a percentage scale using daily instead of monthly stats, it should be easier to identify the volatile periods. How do I do that in Excel? What charting app should I use?
That huge spike in April 2005 (1209 comments) is clearly the work of our friends at
wearesphinx. It should be interesting to relate other spikes with specific events:- Within the community (above mentioned friends and their brethren),
- Within LiveJournal (invite codes removed Dec 2003, spike from 81 to 236 comments in Jan 2004, remains above 100 henceforth), or
- Elsewhere (Rajkumar dies, riots in Bangalore Apr 2006, 535 comments).
