Data Mining Model Training Destination

The Data Mining Model Training destination trains data mining models by passing the data that the destination receives through the data mining model algorithms. Multiple data mining models can be trained by one destination if the models are built on the same data mining structure. For more information, see Mining Structure Columns and Mining Model Columns.

Configuration of the Data Mining Model Training Destination

If a case level column of the target structure and the models built on the structure has the content type KEY TIME or KEY SEQUENCE, the input data must be sorted on that column. For example, models built using the Microsoft Time Series algorithm use the content type KEY TIME. If input data is not sorted, the processing of the model may fail. If the data requires sorting, you can use a Sort transformation earlier in the data flow to sort the data. This requirement does not apply to columns with the KEY content type. For more information, see Content Types (Data Mining) and Sort Transformation.


The input to the Data Mining Model training destination must be sorted. To sort the data, you can include a Sort destination upstream from the Data Mining Model Training destination in the data flow. For more information, see Sort Transformation.

This destination has one input and no output.

The Data Mining Model Training destination uses an SQL Server Analysis Services connection manager to connect to the Analysis Services project or the instance of Analysis Services that contains the mining structure and mining models that the destination trains. For more information, see Analysis Services Connection Manager.

You can set properties through SSIS Designer or programmatically.

For more information about the properties that you can set in the Data Mining Model Training Editor dialog box, click one of the following topics: