Building my first model

Discussion in 'microsoft.public.sqlserver.datamining' started by Kastellu, Apr 1, 2007.

  1. Kastellu

    Kastellu Guest


    I hope this newsgroup is kind of free to ask whatever and whenever i need to know about datamining , because i am starting with my first data mining model in SQL server 2005 and i will have lots of questions.

    I have two tables one is a item dimensional table and the second one is fact with inforamtions about good peaces and bad(scrap) pieces or units of thouse products. I can stick to it many other dimensional tables , like table that tells me the type or reason why te unit is bad, date dimension, ...

    I would like to predict what is the chance that the product will be bad and what the reason is(by reason).
    Of course if you have other suggestions please tell me , this i just an example.

    The first problem that i had was chousing the right data mining technique, what do you sugest?
    What should be the input and what should be prediction? Key (ItemID), Input(ScrapQuantity), Prediction(ScrapID<-- reason or type fault)?

    That is if for start.
    Thank you for your help.

    Best regards,
    Kastellu, Apr 1, 2007
    1. Advertisements

  2. Kastellu

    Dejan Sarka Guest


    You can use one dimension as the case tablee and fact table as the nested
    table. You can use Decision Trees and Naive Bayes algorithms for a
    beginning. Then you can use ScrapId as predictable column, and any other
    non-key columns from the dimension as input columns. If you need attributes
    from multiple dimensions, this can be a bit tricky. Check how to do it at
    Dejan Sarka, Apr 3, 2007
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.