SQL Server Developer Center > SQL Server Forums > Data Mining > Efficient way to cluster points on line y=ax+b? and now extend the idea to higher dimensional data?
Ask a questionAsk a question
 

AnswerEfficient way to cluster points on line y=ax+b? and now extend the idea to higher dimensional data?

Answers

  • Friday, August 17, 2007 5:32 PMShuvro MitraAnswererUsers MedalsUsers MedalsUsers MedalsUsers MedalsUsers Medals
     Answer

    Are you using SQL Server Data Mining for clustering?

     

    For points (x,y) in the line y=ax+b, you'll detect clusters only using attribute x since y is a dependent variable. Any clustering algorithm will be able to handle this efficiently. When you mention higher dimensional data, I assume you mean more independent variables, but the idea is the same.

     

    Please let me know any specific questions you have regarding SQL Server Data Mining or Clustering algorithm in general

All Replies

  • Friday, August 17, 2007 5:32 PMShuvro MitraAnswererUsers MedalsUsers MedalsUsers MedalsUsers MedalsUsers Medals
     Answer

    Are you using SQL Server Data Mining for clustering?

     

    For points (x,y) in the line y=ax+b, you'll detect clusters only using attribute x since y is a dependent variable. Any clustering algorithm will be able to handle this efficiently. When you mention higher dimensional data, I assume you mean more independent variables, but the idea is the same.

     

    Please let me know any specific questions you have regarding SQL Server Data Mining or Clustering algorithm in general

  • Wednesday, July 01, 2009 6:17 AMGuennadiy Vanine Users MedalsUsers MedalsUsers MedalsUsers MedalsUsers Medals
     
    For points (x,y) in the line y=ax+b, you'll detect clusters only using attribute x since y is a dependent variable. Any clustering algorithm will be able to handle this efficiently. When you mention higher dimensional data, I assume you mean more independent variables, but the idea is the same.

    Please let me know any specific questions you have regarding SQL Server Data Mining or Clustering algorithm in general

    Can this be described in more detail?

    I could not grasp who will do what and how. That is - to discern between "you'll detect" and "Any clustering algorithm will be able...".

    Also, I could not understand why "only using attribute x since y is a dependent variable".
    To me, this is quite symmetric:
    x is also dependent on y as x = (y - b) /a


    Guennadi Vanine -- Gennady Vanin -- Геннадий Ванин