I suggest you ...

add data mining features

104 votes
Vote
Sign in
Check!
(thinking…)
Reset
or sign in with
  • facebook
  • google
    Password icon
    I agree to the terms of service
    Signed in as (Sign out)
    You have left! (?) (thinking…)
    anonymous shared this idea  ·   ·  Flag idea as inappropriate…  ·  Admin →

    5 comments

    Sign in
    Check!
    (thinking…)
    Reset
    or sign in with
    • facebook
    • google
      Password icon
      I agree to the terms of service
      Signed in as (Sign out)
      Submitting...
      • Vague S. commented  ·   ·  Flag as inappropriate

        regressions can absolutely be used for data mining and are one of many data mining tools. i don't think he was suggesting that regressions are data mining, he just wanted a few tools for data mining which certainly do include regression capabilities. if you can't even run a regression, how do you know if the model is appropriate or not?

      • anthony.brockwell commented  ·   ·  Flag as inappropriate

        Including regression (i.e. standard linear regression, including multiple regression), and logistic regression is a very good idea. They are both very useful tools, and relatively easy to code up.

        But PLEASE DON'T call this "data mining" in the final release. Regression is a statistical analysis tool that can be used (or abused) in the higher-level process of what some people call "data mining". Data mining is the process of searching through many possible models in an attempt to find relationships; this is different from carrying out a regression.

      • aeblank commented  ·   ·  Flag as inappropriate

        I think data mining would be a great addition -- particularly regressions. I realize that regressions can be implemented by solving for a system of equations, but that is much more complicated to implement, and can get nasty for non-trivial regressions. I posted about this in the forum, and found a few commercial packages: http://www.extremeoptimization.com/quickstart/PolynomialRegressionCS.aspx and http://www.kdkeys.net/forums/thread/605.aspx, as well as a free package at http://www.kdkeys.net/forums/thread/605.aspx but it doesn't seem to be very powerful (the free package), nor would it integrate as nicely with the rest of Iridium.
        Wish I could be of more help, but I'm well versed in the theory behind regressions and am used to using stats packages to do statistical analysis, but would love to be able to access functions like "linear_regression" or "logistic_regression" from within c# in a clean, well-programmed way!

      Feedback and Knowledge Base