I am getting this error: Error (Data mining): The INSERT INTO statement can be used only for mining models that have no siblings in the same mining structure. Try using INSERT INTO Mining Structure instead

Proposed Answer I am getting this error: Error (Data mining): The INSERT INTO statement can be used only for mining models that have no siblings in the same mining structure. Try using INSERT INTO Mining Structure instead

  • Thursday, October 27, 2011 6:29 PM
     
     

    I am issuing a retraining dmx query in ssas 05 in order to retrain a specific model.  The query looks like this below.  When I execute it I get this message.  How should I change the query so that it works?

    Error (Data mining): The INSERT INTO statement can be used only for mining models that have no siblings in the same mining structure. Try using INSERT INTO Mining Structure instead

     

    Insert

    Into MM_Model1 (

    [CustomerID]

    ,[Age]

    ,[ItemsBought]

    )

     

    OPENQUERY([MyDatabase],

    'SELECT

    [CustomerID]

    ,[Age]

    ,[ItemsBought]

    From MyTable'

    )

All Replies

  • Thursday, October 27, 2011 7:27 PM
    Moderator
     
     Proposed Answer

    If your structure has data (other trained models) you only need to do

    INSERT INTO <ModelName>

    The structure data will be used.

    Otherwise, use

    INSERT INTO MINING STRUCTURE <structureName>(

    [CustomerID]

    ,[Age]

    ,[ItemsBought]

    )

     

    OPENQUERY([MyDatabase],

    'SELECT

    [CustomerID]

    ,[Age]

    ,[ItemsBought]

    From MyTable'

    )


    bogdan crivat / http://www.bogdancrivat.net/dm
  • Friday, October 28, 2011 11:20 PM
     
     

    My understanding is that I can delete data from specific mining models or specific mining structures. Is that correct?  Is it possible to delete data from mining models but not from the structure?

    If I execute this command: INSERT INTO MINING STRUCTURE <structureName> Does it train all the models in the structure?  How would I train only specific models?

     

  • Monday, October 31, 2011 3:26 AM
    Moderator
     
     Proposed Answer

    DELETE  FROM MINING MODEL <a> -- untrains the model. Learned patterns are gone, data stays in the parent mining structure

    INSERT INTO MINING MODEL <a> -- trains the model using data in the structure (if any) -- it will fail if the structure is unprocessed. Other models are untouched

    INSERT INTO MINING STRUCTURE <ms> --the equivalent of ProcessDefault on the mining structure -- takes the shortest path to fully processed object and descendants. If the structure has data, it will process unprocessed models. If the structure has no data, but it has bindings (i.e. created in BIDS), it will process the structure and then the models. If no data is available, and no bindings, then an error will be raised

     

    Hope this helps

     

     

     


    bogdan crivat / http://www.bogdancrivat.net/dm