KPG Activity Measured by Posts per Hour

Data and chart: http://i.imgur.com/hvRZzz1.jpg
Line graph: http://i.imgur.com/RATkhOw.jpg (Note that I wrapped the line graph to show continuity from 11:00pm EDT to 12:00am EDT the next day)

All times are in EDT, which is GMT -4. The data comes from 12 most recent threads excluding the Halloween thread because DST changed in the middle of that thread which would have skewed the data.

The average number of posts made per hour is 59. During “very active” periods (in red), that figure increases to 87 posts/hour. During “inactive” periods (in blue), the average drops to 37 posts/hour. 25% of all posts are made in a four hour period between 12:00pm EDT and 4:00pm EDT. By comparison, during the least active four hour period, only 9% of posts are made. The fluctuations here are not insignificant; from lowest to highest periods, activity seems to slightly more than double.

The most active hour is 1:00pm EDT; the least active hour is 5:00am EDT. These times match well with the statistics mentioned above and the color summary shown in the above chart. As the color summary shows, there is a noticeable cool down - heat up trend that occurs cyclically. Early hours are marked by low activity; as early morning moves to late morning activity increases; by afternoon to mid late afternoon activity is at peak; evening to nighttime we are somewhat stagnant with pockets of higher activity; and as night turns into the next morning activity decreases again.

The time with the greatest standard deviation, indicating more range of the data, is 10:00am EDT. This is in part caused by a huge abnormal jump in activity in that hour on one day in early October. It’s unclear what occurred that day at that time to cause the large jump. With 253 posts the thread in question also happened to contain the greatest posts per hour recorded (keep in mind the average is 59 posts/hour). In contrast, the least posts per hour recorded in a thread was 9 posts at 4:00am EDT in a thread in mid October.

I tested the correlation between the individual data from each thread and the overall average and consistently came up with strong positive correlative results ranging from .54 to .78. This means we can trust the data to provide reliable results from which we are able to make conclusions like the ones above.