Stats on Posts/Day (or...geeking out with ggplot2)

[ECO]

As I was making my previous post about the 20,000 post milestone, I was interested in the real answer for how many posts/day we have had over the life of the site.

Unfortunately vBulletin has only very rudimentary stats built in. As in one text column of posts per day. Dropped the data into R to get some prettier representations...and it looks like we're >60/day for the last full month:

Then I realized this included the typically slow weekends...hence the huge middle 50%...a bit of filtering later shows we're over 70/day during the week:

Splitting the data up by day of the week shows what one might expect, Tues-Thurs are the highest activity with some drop off for Mon/Fri:

If nothing else, I hope this post encourages you to explore the R/ggplot2 combo if you haven't already.

setwd('/Users/eco/Documents/WWW/SEQanswers.com/stats')
d <- read.csv('stats.csv')
days <- c('Sun','Mon','Tues','Weds','Thurs','Fri','Sat')
d$realday <- days[d$weekday]
d$realday <- factor(d$realday, levels=days)

png("monthly.png",width=750,height=350) 
qplot(factor(month),posts,data=d,colour=factor(year),geom='boxplot',ylab='Posts/Day',xlab='Month',main='SEQanswers.com Post History (all 7 days)')
dev.off()

png("monthly_weekdays.png",width=750,height=350)
qplot(factor(month),posts,data=d[d$weekday>1 & d$weekday < 7,],colour=factor(year),geom='boxplot',ylab='Posts/Day',xlab='Month',main='SEQanswers.com Post History (Weekdays Only)')
dev.off()

png("weekdays.png",width=750,height=350)
qplot(factor(realday),posts,data=d,colour=factor(year),geom='boxplot',ylab='Posts/Day',xlab='Weekday',main='SEQanswers.com Post History by Weekday')
dev.off()

[bioinfosm]

Excellent Eco.. shows that Wed is not that sleepy after all! Not for NGS :P