Using the data at the following link I did a little time series analysis of the monthly archived KBs for the Programming forum from October 2005 to July 2008 [because the first and last months both looked as if they were only partial months I excluded both]. I did the analysis by copy and pasting the data 33 rows only from the following link into a new .ijs file and adding a line at the beginning and at the end as suggested by the excerpt shown below (beware of line wrap).
http://www.jsoftware.com/pipermail/programming/ ******begin exerpted temp file********** programming =: 0 : 0 July 2008: [ Thread ] [ Subject ] [ Author ] [ Date ] [ Gzip'd Text 27 KB ] June 2008: [ Thread ] [ Subject ] [ Author ] [ Date ] [ Gzip'd Text 124 KB ] [snip] December 2005: [ Thread ] [ Subject ] [ Author ] [ Date ] [ Gzip'd Text 142 KB ] November 2005: [ Thread ] [ Subject ] [ Author ] [ Date ] [ Gzip'd Text 86 KB ] October 2005: [ Thread ] [ Subject ] [ Author ] [ Date ] [ Gzip'd Text 14 KB ] ) ******end exerpted temp file********** I then loaded the plot and stats scripts and renamed the data set as follows. load 'plot' load 'stats' load '/Users/brian/j602-user/temp/13.ijs' $programming 2682 $p =: programming 2682 $<;._2 p 34 The following line plots all of the Programming data. 'xtic 12' plot |. }:}.". @((_5}.(>:@i:&'t')}.]));._2 p The following line plots a 12 month centered moving average of the Programming data for which the first 6 and last 6 months are elided. Such a CMA is often suggested as a simple way to eliminate any seasonalities in time series data. 'xtic 12' plot -:((12 median \])@}:+(12 median \])@}.)|. }:}.". @((_5}.(>:@i:&'t')}.]));._2 p This last plot seems to suggest that there was a step up in message volume about x=7+6 to the present, but there may have been a slight decline at about x=17+6 . I have added 6 months to each x value because the centered moving average eliminated 6 months at both the leading and trailing ends. So if x=0 corresponds to November 2005, I figure x=7+6 corresponds to December 2006. A similar analysis could presumably be done readily for the General archive, but I am not sure how informative any of this is. -- (B=) ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
