Announcement

Collapse
No announcement yet.

Stats hard to justify

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #46
    Originally posted by JohnSunnyhills View Post
    I notice NZAA8 and NZAA22 are both registering uptime of arount 716/717. Both in ths same city to me. Why introduce compensation where clearly inthe right location with the right gear you can achieve a good ranking. What gear are they using?
    You make a good point. Perhaps the thing to do is to aggregate the report count for a given airfield in the compensation factor.

    Alternatively, they are getting some high altitude traffic that you're not, although even Northland isn't exactly on the way to anywhere. Or else they're seeing some low altitude local traffic that you can't. FR24 probably doesn't care what's being reported, so long as something is being reported for the purposes of uptime.

    ETA: You would not expect Invercargill to get much traffic but search for NZNV on shared stats and F-NZNV3 reports 716 hours of uptime.

    Hrm. This warrants further investigation, but only FR24 have the data to do so.

    ETA 2: Uptime is not a substitute for range in terms of utility. Just because somebody sees farther than you is not a reason to advantage them in terms of uptime. That's why there are other terms in the score formula to allow for range and such.
    Last edited by Strix technica; 2017-05-07, 00:06. Reason: add invercargill; add uptime utility.

    Comment


    • #47
      Originally posted by MrMac View Post
      The score should reflect the parameters that a feeder can change, improve, and also see on the individual stats page. If the range is poor, it will already be reflected in the stats as it reduces both max and avg range. If I provide a bad internet connection, or turn the box off at night, it should be visible in the uptime stats. But, if I provide power and internet 24/7, and always get 100% uptime in the individual stats, the public stats SHOULD show 720 hours.
      ^ This.

      Originally posted by MrMac View Post
      just change the uptime to mean the same as the 100% individual uptime
      So how would you prevent bad actors from gaming the system? Wherever there are league tables, there are those who try to get to the top by hook or by crook because they can.

      All I need to do is point fr24feed at the discard port (9) and I don't even need an ADS-B decoder running to get 720 hours which, as I've observed elsewhere, is weighted at least 130 times realistic max range (the current next most significant term) — at least when my real receiver isn't available. It's an example of the Hawthorne Effect; observation changes behaviour.

      Originally posted by MrMac View Post
      That would make a much more transparent stats calculation without any complicated compensation formulas.
      Agreed. Such compensation factors are weak hacks at best, but I proposed it on the assumption that FR24 require some way of preventing people from gaming the system. At least with such a compensation factor, you'd need to have some valid traffic during the period when the compensation factor was in effect to get anywhere.

      Also, I can't model its effect without knowing more about FR24's infra and other details. So it's a first rough draft.
      Last edited by Strix technica; 2017-05-07, 10:59.

      Comment


      • #48
        Originally posted by JohnSunnyhills View Post
        As a feeder you only need basic stats as you say. I am thinking do IU buy/setup a new antenna, do I apply for a internet fibre service to be installed. the simple stats assist here.
        Again, you make a good point, but the public stats are part of the incentive to improve your station.

        Also, range is a poor metric of quality. If I can develop a better metric, maybe FR24 might consider using it.

        Originally posted by JohnSunnyhills View Post
        do I apply for a internet fibre service to be installed
        With Telecom? You've got to be kidding if you live in a remote, rural area. It's hard enough to get broadband to everywhere in the UK even with OfCom breathing down BT's neck, and it's not like BT can't afford it. It'd be cheaper and easier to use a mobile service, or even use Inmarsat or Iridium (ADS-B messages are hardly bandwidth intensive) than get Telecom to spend so much as a penny unless they think they can make decent coin out of it.

        (Oops, don't have cents in NZ anymore. 10¢ was the smallest coin last I knew. And they've replaced them all although, in fairness, the old 50¢ coin was kind of ridiculously large and heavy. We're due for new coinage and plastic notes in the coming years. Already got a plastic fiver. I would say pennies and tuppence were pointless, but who knows given the current political climate... we might be back to farthings in a few years.)

        This is going back a bit and it's probably no longer true, but many new real estate developments in NZ couldn't get broadband because their telephone service was provided by fibre to a peripheral CO without a DSLAM, and Telecom wouldn't install a DSLAM unless enough customers demanded ADSL.

        Originally posted by JohnSunnyhills View Post
        All the other ones are academic
        You say that like it's a bad thing
        Last edited by Strix technica; 2017-05-07, 12:17.

        Comment


        • #49
          Just a question on weird results. At NZAA there is now a fleet of ground vehicles setup with transponders. Do the signals from them count as activity. If you have line of sight to the runway you can get your hours up by vehicles picking up rubbish off runways all night. This is probably the case at the majority of airports now.

          Comment


          • #50
            Originally posted by JohnSunnyhills View Post
            At NZAA there is now a fleet of ground vehicles setup with transponders. Do the signals from them count as activity. If you have line of sight to the runway you can get your hours up by vehicles picking up rubbish off runways all night. This is probably the case at the majority of airports now.
            Good question. Probably not, assuming they're set up correctly. If the SBS feed description is anything to go by, there are separate message types for surface and airborne positions.

            Comment


            • #51
              On the question of whether public stats should show availability vs hours uploaded, I've had a few more thoughts.

              There are several radars close enough to me to be named after the same airfield, so it follows (given the density of airfields around here) that we have similar views of the sky, and yet two of these radars have 708 and 714 hours but not 720 hours according to the public stats.

              I haven't been feeding for 30 days yet so I can't be sure what my 30 day uptime will look like yet, but on the few occasions I've looked at the differences in stats from day to day, it's always increased 24 hours on the previous day. Also, my own data show that there has never been a day, since I started collecting these data, without at least 1 aircraft with position in sight in any given 5 minute period at any time of the day, usually more like 4-6 (especially since I improved my rig).

              Assuming their receivers were up 24x7, this would suggest that they are not seeing all the aircraft I'm seeing. That is significant in terms of utility although, as someone pointed out, uptime should mean uptime, and observed a/c should be accounted for separately.

              The public stats don't give any account for the number of a/c seen, only average and maximum range. There's good reason for that: those located near international hub airports are going to see many more a/c than those located near a regional airport that is closed at night or else not located near any airport at all.

              Still, the minimum number of distinct aircraft might be useful for accounting for time where fr24feed is connected but reports no messages. Example: make uptime the number of hours for which fr24feed was connected (regardless of messages relayed) and introduce a negative term that is the number of hours connected with zero messages relayed. This would be equivalent to the current scoring formula but give greater transparency and it puts the uptime figure squarely in the hands of radar operators.

              In order to make things fairer on those for whom there really are few (or no) a/c at night, a compromise might be to use bigger bins for "hours up with messages". At the moment, the bin size seems to be 1 minute, but that could be expanded to 1 hour at night at that longitude. A radar that reports only one a/c per hour is almost certainly up and continuously able to report a/c, there just aren't always a/c to report. They oughtn't be punished in the rankings for there being nothing to report.

              This is less ambitious (and therefore perhaps more palatable to FR24) than the cosinusoidal compensation formula I proposed the other day and though it might not completely solve the problem for the likes of Oblivian and JohnSunnyhills, it would go a long way to mitigating it.

              Going back to my nearby radars: both report higher average range (although this may change after 30 days' worth with my current rig) and one reports a higher max range. That guy probably has his antenna outside, where mine is in the loft.

              Either both of these guys actually experienced some downtime in the last 30 days and none of the above would help them (but their uptimes got me to thinking with the results described above), or their pattern is different to mine (eg because of local obstructions). Mine is biased toward the east which means I probably have a better chance of seeing nocturnal traffic than others biased toward other sectors.

              Comment


              • #52
                Just noticed changes to the formatting on the Individual Statistics pages this morning - and some additional data provided ...

                ylis

                Comment


                • #53
                  Just wanted to make a quick post here that more stats are going to be added to individual stats page soon. And when that's done, we will try to take a look at the share stats page to see if the formula could be tweaked to make it more fair.
                  --

                  Comment


                  • #54
                    Originally posted by Khan View Post
                    Just wanted to make a quick post here that more stats are going to be added to individual stats page soon. And when that's done, we will try to take a look at the share stats page to see if the formula could be tweaked to make it more fair.
                    Khan

                    Thank you for the update. I can't even pretend to understand the formula discussion, it's way over my head but from the stats page itself can see that it's worth a tweak or two. Hope the 'try' can turn into a 'will do.'

                    Regards,
                    Gregg
                    YSSY2/T-YSSY4 [SBS-1 Basestation w/- SSE-1090 SJ Mk2 Antenna (Thanks Delcomp) ] [Uniden UBCD996T w/- 16 element Wideband Discone VHF/UHF Antenna, and tuned 108MHz-137MHz Airband Antenna] [Trialing a home-brew 1090MHz collinear antenna]

                    Comment


                    • #55
                      Originally posted by fungus View Post
                      Khan

                      Thank you for the update. I can't even pretend to understand the formula discussion, it's way over my head but from the stats page itself can see that it's worth a tweak or two. Hope the 'try' can turn into a 'will do.'

                      Regards,
                      Gregg
                      Gregg

                      The first priority is updating of individual stats page so when that's done then will be Share Stats. The reason why I wrote "try" was because I didn't want to emphasize that it was going to be done right away
                      --

                      Comment


                      • #56
                        Originally posted by Khan View Post
                        Gregg

                        The first priority is updating of individual stats page so when that's done then will be Share Stats. The reason why I wrote "try" was because I didn't want to emphasize that it was going to be done right away
                        Copy that and understand.

                        Regards,
                        Gregg
                        YSSY2/T-YSSY4 [SBS-1 Basestation w/- SSE-1090 SJ Mk2 Antenna (Thanks Delcomp) ] [Uniden UBCD996T w/- 16 element Wideband Discone VHF/UHF Antenna, and tuned 108MHz-137MHz Airband Antenna] [Trialing a home-brew 1090MHz collinear antenna]

                        Comment


                        • #57
                          Originally posted by JohnSunnyhills View Post
                          Just a further thought..
                          I notice NZAA8 and NZAA22 are both registering uptime of arount 716/717. Both in ths same city to me. Why introduce compensation where clearly inthe right location with the right gear you can achieve a good ranking. What gear are they using?
                          Maybe they are enjoying panoramic views from homes on the uppermost slopes of One Tree Hill. I would be interested to know these answers too.
                          This was a very interesting thread back in April.
                          Since then I have been searching for answers as to how to increase range and uptime.
                          I contacted lokeynz who runs the top NZ radar NZWS2 which has an excellent range 310 and uptime of about 685. He sent me a photo showing his radar was 1200 metres high on top of a ridge in the Southern Alps on a comms tower.
                          Several weeks ago the uptime started to reduce, I contacted him again and found that he was shutting down the radar at night in order to preserve the solar powered usage, necessary because of mist and clouds this time of year.

                          In this thread I suggested that my poor uptime was due to Auck Airport shutting down in early morning, that turned out tobe false.

                          I followed the guidelines on the top of the stats page, 1 high aerial, 2 short cable 3 raspberry 24/7
                          I was already doing 3
                          I setup a couple of watertight boxes and relocated my two raspberrys onto the roof, decided to run cat6 rather than rely on wifi. Distance to aerial was 2 metres.
                          I purchased a new antenna from jetvision.de (A3 ADS-B) and attached it to NZAA4. NZMB22 is on my old antenna.

                          Since then NZAA4 has increased uptime so it is currently up for 718 hours. NZMB22 is back at 687 (by the way NZAA4 is a Beast, NZMB22 is a DVBT stick which is also running mlat)
                          So my current feeling is that the stats are useful to use when making changes, don't mean a lot in themselves.
                          Depending on your location changes can be made that affect performance.

                          As an aside there was a period earlier this year when I placed a 3 metre mast on top of my roof and a signal amp down 20 metres of cable and got the range to 185 using my old antenna. Uptime was still around 690. So height affects range but cable length became a factor in signal quality.
                          Last edited by JohnSunnyhills; 2017-08-16, 21:46.

                          Comment


                          • #58
                            @JohnSunnyhills:
                            Yes, you got the right approach and right conclusion. It is good commercial antenna, placed at a location where it can have un-obstructed view of horizon in most directiond, and a short coax which improve the stats. Feeding 24/7 by using a Pi is also a big help in improving stats, and is a low-cost + headache free solution.

                            Increasing height of antenna helps in overcoming obstructions caused by surrounding houses, tall buildings, and trees. Raising antenna beyond overcoming the obstructions to clear view of horizon, does not help much. Beyond this point, it is the terrain which decides the maximum range.

                            What is the Maximum Range I can Get?

                            If you want, you can monitor performance of your system, and results of changes by monitoring data locally at your Pi by installing Performance Graphs.

                            Please see this thread for details:

                            Automated Installation of Dump1090-Mutability, Data Feeders, and Performance Graphs

                            .
                            Last edited by abcd567; 2017-08-16, 22:40.

                            Comment


                            • #59
                              Has anyone worked out ( or does anyone know ) what time period the Shared Statistics relate to ? These seem to be refreshed at about 1000UTC but can be as early as 0700UTC on some days. Do they cover 0000UTC-2359UTC ?? or something different ??

                              ylis

                              Comment


                              • #60
                                I'd like to know, what is the point of the ranking, I would consider to remove it. In our community that feed radiosondes data, we can do without it. Sending data is quite enough.

                                Comment

                                Working...
                                X