Creating Relational Data Mining Models
The Mining Model Wizard is used to create a relational data mining model. For more information about the Mining Model Wizard and the steps needed to create a relational data mining model, see Mining Model Wizard.
The Mining Model Wizard uses the case tables, data mining technique, case key column, input columns, and predictable columns you provide to build and optionally process a mining model.
The case tables contain the columns needed to establish a case set for a data mining model. In the case of a single case table, a single relational database table contains all of the information needed. With multiple case tables, however, the information may be distributed across several tables and joins between relational database tables are needed to establish the case set.
When you select a data mining technique, you choose a data mining algorithm to be used with the relational data mining model. By doing this, you implicitly select a data mining algorithm provider. The data mining algorithm provider supplies the data mining algorithms and dictates the model structure for the data mining model.
You also select input and prediction columns when you use the wizard to create a relational mining model. Input columns are used by the data mining provider to contain training data. Selection columns are used to provide predictive analysis results when querying a data mining model.
Prerequisites
Before you build a relational data mining model, a relational database containing a table structure for training data must exist, so that the Mining Model Wizard can use the table structure to define data mining columns. The Mining Model Wizard can also train the data mining model. Relational data mining models cannot be created from OLAP data sources.