Various graphs and unavailable info

1974ER · Post by **1974ER** » Mon Nov 10, 2008 4:28 am

Hello everyone!

I have been taking a look at various graphs. In many cases having 0 entries causes "interesting" things to show up... or rather not show up. For example, if there are very few snapshots to base daily activity graphs on, a graph may only cover part of the day. Part of this seems to be related to the fact the graphs appear to need at least one entry at both 23 and 0 hours to cover the whole day.

I am not sure, if it is possible to do the following, but it should increase overall info accuracy a lot. Set 1 entry of 0 to each of the 24 hours. This would mean that a server that does not get any reports, would show 24 entries of 0 per day... and get a very accurate "no activity reported" at the end of the month, because all servers would get an absolute minimum of 28(occasional 29)/30/31 x 24 entries each month. If there is a player report, it replaces the entry of 0. If there are multiple player reports for the same hour, they get added exactly as they do now.

In effect: Servers with no player made reports would show lines of 0. Servers with just 1 player made report would show a single spike and extremely low overall activity value. Servers with multiple player made reports would show reports very similar to the ones appearing now, but with 24 hour graphs and in cases where some timeperiod has no data, slightly lower overall activity.

As things are now, there are lots of servers, which have red figures and effectively 0% accuracy, many graphs are "broken" and in some cases the info ends up being very strange.

An example is EU-Ahn'Qiraj Realm, which as of this writing has one (1) Alliance entry of 670 and 141 Horde entries with an overall of 602. This means that according to completely unreliable Alliance data this server has much higher Alliace activity than Horde activity (which is very accurate for the time period of 17 to 8, but totally questionable from 9 to 16, because that period has NONE of the 141 entries).

The Alliance side has no graph and shows a flat average of 670. The Horde graph is "full", because it luckily has several 23 and 0 hour entries... but... it show a steady increase between 8 and 17, which can not be right, because a) players come and go and b) there is noone actively making snapshots.

If the changes I suggested were possible... Alliance would have a spike of 670 at 20 hours and overall activity of 28... and the Horde graph would look a bit different, but due to the hole in entries, overall activity would go down to 401.

Just because I curious, I am going to quit writing now and see if I can snapshot both sides to make 11 entries and then wait for the effects to be processed and entered and then look at the graphs again.

Post by **Rollie** » Mon Nov 10, 2008 9:54 am

I'll have to take a peek at what you are referring to and see if I can get a better grasp of what you are suggesting.

1974ER · Post by **1974ER** » Mon Nov 10, 2008 3:14 pm

Hello Rollie!

English is not my native language and I was quite tired when wrote the post above (slept poorly because my stomach was upset), so maybe I did not manage to express myself clearly enough. And I made slight error speaking about months, because you are actually tracking last 30 days, day by day.

I will try another couple of examples, this time purely made up to make it easier for me to count.

Server X daily info has 1 Alliance entry, 2400 players at 12. Horde has 120 entries, 5 for each hour, avaraging 1200 players overall, with reported high of 2400 at 18. Now... Alliance has no solid blue graph, the blue dotted line is at 2400. Horde has a solid red line going up and down some and a dotted red line at 1200. So, even though we have very good info for the Horde... the lack of Alliance data causes the overall picture to look like as Alliance had a 2:1 activity rate compared to the Horde.

Now, someone takes snapshots of the Alliance side at 6 and at 18, and gets 2400 players both times... The graph changes as follows: there is now a solid blue line starting at 6 and ending at 18, with dotted lines going from 0 to 6 and from 18 to 24. We have more Alliance data... which should be good... but it is not, not really... Because it suggests that the Alliance has a very steady population of 2400. The following day, Blizzard decides to shutdown the server for emergency maintenance at 5 and due to major problems it lasts till 19. The person who did the Alliance snapshots does not get online at all. The Horde players swarm the server in far greater numbers than the Alliance, but do not manage to exceed the 1200 average nor the high of 2400... several of them make snapshots and the information gets entered here... Result... The Horde players get "punished" for being active, because they now have even less average activity than earlier.

If no info was treated as an entry of 0... Horde graph would not change at all in the above scenario day 1... during day 2 it would change according to the info sent by Horde players just as always. But for the Alliance... The miserable amount of info would look like this on day 1: Solid blue line from 0 to 11 at 0, then going up to 2400 at 12 and coming back down to 0 at 13 to continue on to 24. The dotted blue line would show an average overall activity of 100. After the 2 snapshots, the Alliance would be "rewarded" with an average activity of 300 and would have spikes at 6, 12 and 18. On day 2 they would funktion as follows... If the info 2400 at 12 is younger than 29 days, no change whatsoever... if it is too old, it gets replaced by 0 at 12 and average overall activity drops to 200... and in both cases... any Alliace player sending a snapshot of activity higher than 304 players at anytime between 0 and 5 or 19 and 24 would help his side and increase our info accuracy even if the number snapshotted was 1.

For you all of this would basically mean the following... Server Y has 25-48 daily entries of 0, red flag both sides, too little information on both. Server Y has 1-24 entries of 0, red flag side(s) with 0(s). Server Y has exactly 48 (24 on both sides) entries of 1+, yellow flag, info is thin, but still fairly reliable. Server Y has more than 24 but less than 96 entries on both sides with no 0s, yellow flag, improvement still needed, but data is ok. Server Y has 96+ entries of 1+ on both sides and no 0s, green flag, very good data. Note that with this division the data reliability could both drop from green to red and rise from red to green without going yellow in between. Also, I picked 96 just because it was a multiple of 24, which might or might not make the checking process easier... You could substitute 100, if it does not make a difference.

Hmmm... I have spent several hours typing this and thinking... hopefully I made no major mistakes, getting tired again... and I still wanted to play some WoW...

Even if you say "No, too difficult/impractical/whatever", thanks for listening.

Hybuir · Post by **Hybuir** » Wed Nov 12, 2008 10:17 am

basically make a visual indicator to show that where we are assuming the population of the server during a period of time rather than actually see them online/strong indications that they are on.