| ethomaz said: Day 1: 10 players / 10 logins Day 2: 30 players / 60 logins Day 3: 20 players / 30 logins Day 4: 50 players / 150 logins * Day 4 is suppose to be a weekend for example * Overall average: 250 logins / 110 players = ~2.2 Sum avg per day: 1 + 2 + 1.5 + 3 = 7.5
if I do overall avg x num of days I have 8.8 but when I sum per day I have 7.5. |
Ahh I see what you meant, they are doing a sum average of the days rather than average over days.
Still a bit bizarre for the sum average to be so far removed from the average multiple, but there's not enough information to map this out but you won't get such discrepencies with a shallow linear fit. An sharp fall-off is pretty normal for MMOs, so that likely explains it but the front would be bloated enough to weigh out the average to look "big". That also explains why they are using a mean over the much more accurate (for this scenario) median.
I suspect that is why we are not given a median, because if the median and average (of 1.8, or average plays per player) were close, then there's no way for the sum and multiple averages to be so far apart. So the average >> median in this scenario. This is a pretty text-book case, actually: https://www.ma.utexas.edu/users/mks/statmistakes/skeweddistributions.html
