Processing Data Mining Objects

Processing a mining structure or mining model differs from processing OLAP objects, such as dimensions and cubes. In OLAP processing you build cubes with aggregations, whereas in data mining processing you create training data and run mining algorithms on that data. For more information, see Processing Analysis Services Objects.

Processing Mining Structures

To control how the models are processed in the mining structure, you can process a mining structure separately from its associated mining models. First, you must process the associated mining structure and then you can process the mining model you are interested in. For example, if you have a mining structure named SalesCluster that has several models associated with it, but you only want to process one model, named SalesCluster_model_1, you can process SalesCluster first, and then process SalesCluster_model_1 by itself. You can process a mining structure using the menus or buttons in the Mining Models tab or Mining Structure tab in Data Mining Editor. In the processing dialog box, select a processing option and click Run.

If you are concerned about conserving hard disk space, note that Analysis Services retains mining structure caches locally. That is, it writes out all the training data to your local hard disk. If you do not want the data cached, you can change the default by setting the MiningStructureCacheMode property on the mining structure to ClearAfterProcessing. This will destroy the cache after models are processed; however, it will also disable drillthrough on the mining structure. For more information, see Using Drillthrough on Mining Models and Mining Structures (Analysis Services - Data Mining).

Also, if you clear the cache, you will not be able to use the holdout test set, if you defined one, and the definition of the test set partition will be lost. For more information about holdout test sets, see Partitioning Data into Training and Testing Sets (Analysis Services - Data Mining).

Processing Mining Models

You must process the Analysis Services models that you define before you can start to work with them. You must also reprocess the mining models whenever you change the mining model structure, update the training data, change an existing mining model, or add a new mining model to the structure.

You can process the mining models in the following ways.

  • Deploy the project. Depending on the project settings and the current state of the project, the mining models in the project can be processed when the project is deployed. When you initiate deployment, processing starts automatically, unless there is a previously processed version on the Analysis Services server and there have been no structural changes. You can deploy a project by selecting Deploy solution from the drop-down list or by pressing the F5 key. You can control how mining models are deployed by setting the deployment properties. For more information about how to deploy an Analysis Services data mining solution, see Deployment (Analysis Services - Data Mining).

  • Use the menus or buttons in the Mining Models tab or Mining Structure tab in Data Mining Editor. In the Mining Models tab, you can reprocess the mining structure, process just the selected mining models, or process all the mining models associated with the structure.

  • Open the Mining Model Viewer tab in the Data Mining Editor. If the mining model structure is out-of-date, the editor prompts you to redeploy and process the mining model.