today for the first time i was really viewing the statistics page. i thought it would count and order to valuable input given to flightaware - but this is not the case - for ranking they just count the number of aircrafts.
so - if one delivers 10,000 aircrafts with each 1 position - he was feeding higher rated value than another with 9,999 aircrafts and 500 positions each. next thing i do not understand is the top user ranking. e.g. for my antenna testing i set up 2 other sites (just temporarily) at nearly the same place. feeding senseless almost the same boring data multiple - sets me ahead of others???
must be another flavor of mathematics than that i’ve learned.
what’s your and FA staff’s opinion - or did i miss something?
No, it doesn’t count false; the numbers are accurate. It may not be counting what you want it to be counting, but that’s a different issue; it has to be ordered somehow! If it troubles you, you don’t have to look at it.
Note that there are separate summaries for users vs. sites if you want to look where your individual site ranks, rather than the total of all your sites.
regarding the ‘test-sites’ i think a special test-site-account’ could make sense. first it would not set others behind me and my ‘double-data-nonsense’ and second maybe it would beware fa from bad data while testing (e.g. aggressive mode)
the value for flightaware is the sum of miles of all aircrafts somebody feeds. maybe it’s not possible to count this. but you can count the number of planes multiplied with the number of positions (up to a maximum value per minute).
So feeding aircraft that are stationary on the ground would not contribute at all to your ranking, and aircraft that are landing / taking off (and are therefore slower) are less valuable? That’s actually backwards in terms of value to FlightAware - coverage of the landing / taking off phases is more important than lots of data for a cruising aircraft.
In terms of counting aircraft * positions, different feeder types produce positions at different rates; are you suggesting that (say) a Planeplotter feed and a piaware feed from the same raw data should be ranked differently?
starting at 20:00 utc and ending at 20:10 utc => 1 aircraft for 10 minutes
maybe he sent 600 positions to fa => result were 600 value-points
but after max-count-per-timeframe (e.g. set from fa to 5 per 60 seconds) => result is 50 value-points
if he had sent just 20 positions in these 10 minutes => result is 20 value-points
if he has a crazy feeder and sent 10000 positions => result is again 50 value-points
Transponders transmit positions at a much lower rate on the ground.
If you are counting “fed aircraft miles” as you suggested, then an aircraft that’s not moving covers no miles.
when counting positions up to a max. value per time-unit - different feeder behaviors shouldn’t matter.
or did i miss something?
Different feeders feed at different rates over time. Unless your max value is something like 1/minute, you will see differences. If you set the max value as high as 1/minute, you’re basically measuring (unique aircraft * time seen) which is a bit different to what I think you had in mind: 1000 aircraft with 10 positions each and 1000 aircraft with 5 positions each would probably actually come out at about the same rank.
Generally, the number of positions is not a good way to compare sites because the rate of positions depends heavily on the feeder software and less on the raw data quality (in general, you see 0.5-2 raw positions/second for everything in range, and then that rapidly drops to zero at the edge of your range, so there is not a huge difference in the raw rate of positions per aircraft even with quite different receivers; all the variation is in the aggregation of the raw data on top of that)
Also, the stats you’d need to do this are not currently collected.
The number of unique aircraft seen is a simple ranking that is easy to understand and doesn’t work too badly. It’s certainly not perfect but I think it is better than what you’ve suggested so far. There are a few corner cases where it’s not very representative (e.g. for aircraft that loiter for a long time) but they’re not too common.
Also keep in mind that “aircraft minutes” is going to be roughly proportional to “rate of aircraft flying through the receiver’s coverage”. “Unique aircraft” is also roughly proportional to “rate of aircraft flying through the receiver’s coverage”. So you may end up measuring something quite similar by a roundabout route.
went out in the garden for a few minutes to reflect. result is: by giving different max. values according to flightlevel you could overcome the aircraft on ground effects acceptable (despite height above ground would be perfect). the value for flightaware would be better represented this way than it is now in my opinion. but agreeing to disagree we both are familiar with i think in a very friendly way!
p.s. but in the end i think as there are so many different feeder types you can’t run the algorithm on the clients and running on the servers would put too much load on the servers for this benefit …
so - count just the planes as it is and establish a switch for live- and test-mode on the sites backend website :)))
FA Team - There is another issue I observed while accessing ‘Positions Reported by Distance from Receiver’ on FA ADS-B Statistics page from my desktop vs mobile. The desktop version shows max range to be 250+ nm while mobile version shows 300+nm. Could be a simple typo on the page rendering. I have tried this with various browsers on both desktop and mobile (Mac/PC, iOS/Android). Can someone please check and fix this.
Maybe there should be a multiplication factor that reflects the uneakness of the of the positions supplied to the system, after all a position that only one station can provide has to be worth more than one provided by 30 or so other sites .