+ Reply to Thread
Results 1 to 5 of 5

Thread: Outlier

  1. #1
    Points: 703, Level: 14
    Level completed: 3%, Points required for next Level: 97

    Posts
    17
    Thanks
    6
    Thanked 0 Times in 0 Posts

    Outlier




    Hi,

    Is there any way to find outlier without having the distribution of data?

    Thanks and Regards

  2. #2
    Human
    Points: 12,666, Level: 73
    Level completed: 54%, Points required for next Level: 184
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,360
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Outlier

    I don't know of any. You must have the data!

  3. #3
    Points: 703, Level: 14
    Level completed: 3%, Points required for next Level: 97

    Posts
    17
    Thanks
    6
    Thanked 0 Times in 0 Posts

    Re: Outlier

    For example any nonparametric test or other method or algorithm?

  4. #4
    Human
    Points: 12,666, Level: 73
    Level completed: 54%, Points required for next Level: 184
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,360
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Outlier

    If you don't have any data it is impossible to do any statistical analysis. You can only do theoretical speculation.

  5. #5
    Probably A Mammal
    Points: 31,087, Level: 100
    Level completed: 0%, Points required for next Level: 0
    bryangoodrich's Avatar
    Location
    Sacramento, California, United States
    Posts
    2,564
    Thanks
    398
    Thanked 618 Times in 551 Posts

    Re: Outlier


    All outlier or anomaly detection is with respect to some model, whether you're looking at a univariate case of observations that deviate from the variable mean (e.g., 2 SD from the mean could be considered a "large" deviation) or looking at those observations that deviate greatly from a fitted trend line. Therefore, it isn't about the distribution to which you think it belongs. It's about what model you want to accept as representing your data and therefore defines what it means to be significantly different from that model. Suppose you make a bad choice of model (choosing means on a non-symmetric distribution of data, for instance). Then you may say a bunch of observations are extreme, to which on that assumption is true, but it is only because that model of the data was a bad choice. The same applies to trying to fit a straight line to a curvilinear relationship. You might classify observations as being extreme, which is true on that assumption, but it's the model choice that was poor. Models depend entirely on what sort of problem you're dealing with, though. How are you trying to model your data? Is it a univariate observation? A multidimensional model? A linear regression? An association rule? In statistical models you'll find literature on outlier detection. In machine learning literature, you'll usually find anomaly detection. They're both dealing with the same fundamental thing, and there are many ways to calculate the deviation from a model, but that is ultimately all you're doing.
    You should definitely use jQuery. It's really great and does all things.

+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats