SQL Server DB and Statistical Analysis

  • Hi Everyone,

     

    I am new to the forum.  Please excuse me if I post the message in the wrong place.

     

    I am working with an engineering firm and using SQL Server database.  The company wants to add some features of trend analysis to the DB.   I understand it is not easy to have it done because it requires both strong programming skills and good math/statistical background.  With the source code for the trend analysis (using Mann-Kendall, Seasonal Kendall, and Related Nonparametric Techniques) available in FORTRAN 77 I am wondering how difficult is to incorporate the statistical functions into the DB.  What issues are needed to consider when trying to develop such program?  I appreciate any expertise opinions and suggestion about the time and cost effective approaches for the needs.

     

    Thank you very much for your time and help.

  • This was removed by the editor as SPAM

  • Difficult but not impossile.  I created a database that does some six sigma analysis of data.

    It took about 4 months to get most of the kinks out..and you are right you need to have some decent mathematical skills.

    The type of difficulty you may have is in implementing statistical features that require interpolation of values that are in statistical table...creating dynamic functions to do integration(needed to determine probability of something is beyond a certain value for comparison of Z values) etc.

    If you need some help getting off on the right foot..feel free to IM me.

     

    --M Kulangara


    Mathew J Kulangara
    sqladventures.blogspot.com

  • thanks, M Kulangara.  I am trying to get a clear picture about the feasibility of this job and let the management makes the decision. 

    Do you think if the FORTRAN source code helps?  What language did you use to program your DB?

    Thanks.

    Autumn1008

  • I coded everything entirely using T-sql. There was some thought about using extended stored procedures with c native code, but the performance gains were minimal( dynamic t-sql functions to do integrals are costly timewise). Fortran source code will likely help you very little other than to give some idea of the business logic required for functions...ie the functions will need to be rewritten in TSQL.

    --M Kulangara

  • It sounds promising but I don't like mathmatical equations especially those statistical equations.  I'll ask you for help if needed.

    Thanks a lot.

    Autumn1008

  • For this thing I have used Mathlab. I am pretty sure that you will get a decent thorugh put with minimal effort with it but be ready to shell out $$$

     


    * Noel

  • Thanks, Noel.  It is certainly easier to have something to base on but the management prefers not to invest $ to have ready-to-use programs.   Instead we take the hard route to program the DB.

    Autumn1008

Viewing 8 posts - 1 through 7 (of 7 total)

You must be logged in to reply to this topic. Login to reply