Identify pattern within a dataset

  • Status: Closed
  • Pris: €275
  • Bidrag mottagna: 16
  • Vinnare: salepi53

Tävlingssammandrag

We have a csv dataset with ~50.000 samples of two analog sampled quantities 'buried' in heavy 'noise'. We are looking for innovative approaches to cancel out the noise/recover 'information'
(samples are 2.5ms apart - an 'event' which is several seconds long happens 4-5 times during sampling. We need to identify WHERE within the dataset, this 'event' happens)

Data available on Dataset_03.zip

Rekommenderade kompetenser

Arbetsgivares feedback

“Spiros was able to provide a PROOF that the patter we were looking for was simply not there and that critical 'pattern edge' detection info needed was lost in noise. Thank you Spiros!”

Profilbild Opticon, Greece.

Topp bidrag från den här tävlingen

Visa fler bidag

Klargörandetavla

  • Opticon
    Tävlingsinnehavare
    • 6 år sedan

    On the 'noisy' data set, you can safely assume, no 'events' during the first 10seconds and the last 10seconds of the data (each row is 2.5ms)

    • 6 år sedan
    1. fahadtan84
      fahadtan84
      • 6 år sedan

      Have you seen my entry? If there is any query please feel free to query.

      • 6 år sedan
  • fareed499
    fareed499
    • 6 år sedan

    I have further processed the data to show that events are consistent.

    • 6 år sedan
  • wdworks
    wdworks
    • 6 år sedan

    Actually i've got the notification that you've commented my entry, but i can't see any comment. Main thing i'm interested is the noise free similar signal image you referred to. So i can check which how effective current filtering is

    • 6 år sedan
  • Opticon
    Tävlingsinnehavare
    • 6 år sedan

    It appears that I can only chat with freelancers who have already filed an entry. If interested please file 'something' so I can send you the additional data

    • 6 år sedan
  • Opticon
    Tävlingsinnehavare
    • 6 år sedan

    it appears that I cannot upload an image here so I will send it to each one of you. It is a noise-free version of the data SIMILAR (not the same) to the one you got, clearly showing four 'events'. This will help getting a better picture on the 'events' size and relative location

    • 6 år sedan
  • Opticon
    Tävlingsinnehavare
    • 6 år sedan

    Answering to some questions: There are no 'duplicate' entries. It just happens that the same value is sampled. Please ignore the last entry (row) missing the second value, it is a side-effect of saving the file. Measurements come from sampling analog values, the ADC is alternating between the two channels sampling column 1 and 2, so yes, in a way, sampling is sequencial

    • 6 år sedan
  • Opticon
    Tävlingsinnehavare
    • 6 år sedan

    Hello and thank you for bidding on this project.

    • 6 år sedan
  • designdeals
    designdeals
    • 6 år sedan

    Need More Details.

    • 6 år sedan
  • wdworks
    wdworks
    • 6 år sedan

    What's up with the last entry with no y value at the end of csv? Is it like sampling was stopped before the y sample was taken. So sampling is sequential and not parallel?

    • 6 år sedan
  • nomimashi
    nomimashi
    • 6 år sedan

    please suggest more details so that i can work on that, what should i supposed to do with this data and there are many duplicates entry also

    • 6 år sedan
  • testevpsa
    testevpsa
    • 6 år sedan

    * What the data represents?
    * How the event interacts with the data? (Increases the value?)
    * What happens after the event?

    • 6 år sedan
  • testevpsa
    testevpsa
    • 6 år sedan

    Please, could you give more details?

    What the data represents?

    • 6 år sedan

Visa fler kommentarer

Hur du kommer igång med tävlingar

  • Lägg upp din tävling

    Lägg upp din tävling Snabbt och enkelt

  • Få massvis med bidrag

    Få massvis med bidrag Från världens alla hörn

  • Utse det bästa bidraget

    Utse det bästa bidraget Ladda ner filerna - enkelt!

Lägg upp en tävling nu eller gå med idag!