Announcement

Collapse
No announcement yet.

Discussion about Flightradar24 database

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

  • #91
    Could someone please let me know the correct discussion for reporting ground vehicles which are showing up as an aircraft icon instead of the grey ground vehicle.
    Thank you.

    Feeding FR24 since 11/09/13

    Comment


    • #92
      Database sub-forum > data errors.

      Comment


      • #93
        Nice one thanks hexfan.

        Feeding FR24 since 11/09/13

        Comment


        • #94
          Some translations from Peru registry regarding the changes someone just submitted. Looks like they too use suffix somewhat frivolously

          -T and -P

          But as no registry lookup we can't tell if the labels on aircraft are official (no spaces) or registered with -. Jetphotos go with - despite no gaps on wing.

          Sent from my EML-L09 using Tapatalk
          Posts not to be taken as official support representation - Just a helpful uploader who tinkers

          Comment


          • #95
            Originally posted by nomad77
            So I asked this question before but didn't see a response.

            Are you saying that when a new aircraft is assigned a previously assigned reg/code, that aircraft information
            should be posted as NEW data and not a change to the existing information. If this is the case do we end up
            with two aircraft sharing the same code in the DB. Should this happen how does the system know which aircraft
            is the current one.
            Unsure. But I believe that is how it ends up being handled. Every change is likely a 'new' entry from his perspective.

            I have no visibility to the structure of the DBase myself to know. I think the entry point Maxi sees is quite basic (hex, airframe details and reg etc) And it's the archived tables that do the magic work when submitted in the background.

            A well coded one, will have a master table with HEX and current in-air flight details being recorded and an auto generated ID.
            Another with airframe details to tie to the hex via lookup and a similarly auto generated ID and timestamp which is cross referenced for every in-air entry and or flight log associated with a separate Key ID (hex).

            I would imagine the data is dumped to an archived table every so often that can then lookup the historic details and using the paired generated ID with changed data timestamp get the different recorded data/variables.
            Posts not to be taken as official support representation - Just a helpful uploader who tinkers

            Comment


            • #96
              Originally posted by Oblivian View Post
              Unsure. But I believe that is how it ends up being handled. Every change is likely a 'new' entry from his perspective.

              I have no visibility to the structure of the DBase myself to know. I think the entry point Maxi sees is quite basic (hex, airframe details and reg etc) And it's the archived tables that do the magic work when submitted in the background.

              A well coded one, will have a master table with HEX and current in-air flight details being recorded and an auto generated ID.
              Another with airframe details to tie to the hex via lookup and a similarly auto generated ID and timestamp which is cross referenced for every in-air entry and or flight log associated with a separate Key ID (hex).

              I would imagine the data is dumped to an archived table every so often that can then lookup the historic details and using the paired generated ID with changed data timestamp get the different recorded data/variables.
              While I am no Guru on data bases I've also understood that each set of records
              needs a unique identifier in order to access that set of records. Since the
              mode S code is unique to one set of records it should be used to access that set

              Using this logic, a new set of records should be created each time a new mode S
              code is entered. Once created any change to the data inside that set of records
              should be made by accessing the records via the unique identifier (mode S code).

              If the data base does not check for duplicates, but simply allows a new set of
              records including a duplicate mode S code then the situation must arise where
              a choice must be made between two or more duplicate records.

              My concern is that if I post information on a new aircraft which uses a previously
              assigned Reg/Mode S, that will be entered as a new record thus leading to
              duplication.

              Ultimately it makes no difference to me where the information is posted, I am
              simply trying to do the right thing.

              Comment


              • #97
                Keys tend to be unique (not allow duplicates), but you can also have searchable indexes. Such as Registration which will also find the entire row from multi tables using joins or AND statements to validate.

                As we have seen it's somehow possible to have 2 airframes with the same mis-coded HEX OR reg, that tends to indicate there is not a single index in use but a number of them/tables.

                If you also take time tags into account, you can to an extent have duplicates and find records based on time order.

                If it is like the local SQL dumping I do, there is a historic table. And there is an in-air table. Each 'leg' of a hex has it's own 'flight' ID. So it can be apparent when searching for a HEX, to see where it did various legs under different registrations over the time periods.

                Similarly here, looking up a registration could conceivably give you all the flights it's done. Including under different HEX or owner. Whereas if you search by HEX, it may only show the current association. (and possibly the reason behind them not associating HEX with region)

                Without seeing what Maxi has to fulfill, it's a big unknown.
                Posts not to be taken as official support representation - Just a helpful uploader who tinkers

                Comment


                • #98
                  Of note is also the description they edited in the first posts. Which depending on interpretation isn't clear

                  Errors:
                  If you find any errors in FR24 aircraft database you can post the errors and correct data here. It could be wrong registration, wrong plane type, wrong airline or any other aircraft data error.

                  New/Added:

                  Every day there are new deliveries and old aircraft getting new owners, so new information have to be added into database daily.
                  Posts not to be taken as official support representation - Just a helpful uploader who tinkers

                  Comment


                  • #99
                    Originally posted by Oblivian View Post
                    Of note is also the description they edited in the first posts. Which depending on interpretation isn't clear

                    Errors:
                    If you find any errors in FR24 aircraft database you can post the errors and correct data here. It could be wrong registration, wrong plane type, wrong airline or any other aircraft data error.

                    New/Added:

                    Every day there are new deliveries and old aircraft getting new owners, so new information have to be added into database daily.
                    I guess I will continue to do it my way until someone tells me not to.

                    Thanks

                    Comment


                    • From my experience of the site the database key is the registration. Which when you think the whole site is about hex codes is a very strange decision..... But the use of external third party data would only be in registration form so I guess that explains the decision....

                      The problem with a badly maintained database is that once a registration is in the DB twice it screws everything and when the third party data also has a registration error the same happens. It’s not about updating it’s about existing errors.

                      It’s a shame the database isn’t managed in a commercial way like the rest of the site, Maxi does a great job, but it needs a database manager to go through take out all the dupes and crap. It would be easy to hold off on updates fix the errors and revise the database. But I know it won’t happen.

                      Comment


                      • Don't think we can make the assumption it's built on Reg without seeing the background architecture. (and likely being laughed at for doing so)

                        I can think of a way where a non indexed/not-unique key could be looked up using joins to produce apparent bad data in a produced output table. For instance...

                        dbasemaybe.JPG

                        If you had 2 current tables merging the stored and live data with unique key, searchable on all aspects, and you look for REG, chances are you get as good as it comes in current form. But if that also did a lookup for associated HEX OR registration. And then further historic join lookups for same record

                        One badly written from/group statement. Boom. The 'current' results you get are not from a single line but merged data of varied age and potentially duplicate/over-lapped

                        I've done it in error on my daily log table by using group by instead of most recent and so on. Big ol mashup of old vs new records leaving a head scratch


                        Guess the point is, we may never know.
                        Posts not to be taken as official support representation - Just a helpful uploader who tinkers

                        Comment


                        • I'm doing research on predicting fuel consumption and flight delay length using some machine learning techniques. My professor suggested I should take a look on flightradar24.com but I cannot find historic data like from 2010-2017 about air flight information like (when, where, number of passenger, type of plane, age of plan ... weather ...) how can I obtain these information ? (I heard from professor that we have to pay for a fee to get these information, it's ok, the problem is data, the cost is not a problem). Thanks, hope some one guide me a little bit.

                          Comment


                          • Originally posted by chickensoup View Post
                            I'm doing research on predicting fuel consumption and flight delay length using some machine learning techniques. My professor suggested I should take a look on flightradar24.com but I cannot find historic data like from 2010-2017 about air flight information like (when, where, number of passenger, type of plane, age of plan ... weather ...) how can I obtain these information ? (I heard from professor that we have to pay for a fee to get these information, it's ok, the problem is data, the cost is not a problem). Thanks, hope some one guide me a little bit.
                            The database for historical logging only really took off after 2012 I think. And for the most part only stretches 365 days

                            The world’s most popular flight tracker. Track planes in real-time on our flight tracker map and get up-to-date flight status & airport information.


                            Sent from my EML-L09 using Tapatalk
                            Posts not to be taken as official support representation - Just a helpful uploader who tinkers

                            Comment


                            • Thank you very much

                              Comment


                              • Got bored.
                                So - Slight improvement to my code to detect HEX out of normal expected range

                                (after a few hrs of frustration attempting to flag 0/Null values which ended up as simple as reporting when the 'Country' was blank..)

                                http://hexlookup.epizy.com/hex.php?search=CBA9A9 (an addition someone posted with Spanish EC- reg)

                                Decodes to binary, flips 1/0s. Converts back to hexadecimal and gives the suggested alternate it *SHOULD* be

                                Vs good result. Which does conversions still, but hides the alternates



                                /edit and I just noticed a spelling mistake and it gets server issues
                                Last edited by Oblivian; 2019-04-19, 10:11.
                                Posts not to be taken as official support representation - Just a helpful uploader who tinkers

                                Comment

                                Working...
                                X