So a friend of mine said that I needed to look at the graph data a bit more closely. I decided to look at a 7 week average (49 days) and a 29 week average (203 days). What I found interesting was how noisy the data was still at 49 days. Here we see a comparison between different EPEL-6 curves using 7, 49 and 203 day moving averages:
Looking at the curves, while EPEL-6 is still "growing" it seems to have plateau-ed in early to mid 2017. A curve by itself is useless, so here is it in comparison to EPEL-7 where the curve for EPEL-7 seems to increase when EPEL-6 leveled out.
 From this, I expect that EPEL-7 will cross over EPEL-6 in mid to late 2018 though with a 203 day graph that would be hard to see. Finally here is a stacked graph of the releases from EPEL-5 onward using 203 day averages.
Stacking this way shows that when RHEL-5 was EOL in 2017, there is an inflection in EPEL-7 growth.

Note: Originally I was going to compare the powers of 7: 7, 49, 343 but I found the 343 to be so smooth it wasn't clear when a change was occurring. I backed it down to the 203 to get some fluctuations.. and then realized that this was close to the standard 200 day moving average that financial organizations use. However, I am not sure they are the same because financial data is usually in 5 day weeks while I am looking at 7 day weeks.

No comments: