Data Mining Columns
Data mining columns are used to define the inputs and outputs used by a data mining model. The data mining column also provides a standard structure against which familiar SQL syntax, such as INSERT for training data and SELECT for predictive analysis, can be used.
The structure and behavior of data mining columns can be viewed and changed by using Relational Mining Model Editor or OLAP Mining Model Editor. In both editors, the structure pane contains the data mining columns used to define the data mining model; the properties for each data mining column, such as data type and content type, can be viewed in the properties pane.
Data mining columns are added to the data mining model at different steps in the Mining Model Wizard, depending on the type of data mining model. For relational data mining models, the Select the key column and Select input and predictable columns steps add the key, input, and predictable data mining columns to the data mining model. For OLAP data mining models, however, three steps are used. The Select case step selects the case dimension and level used to create key data mining columns, the Select the predicted entity step creates the predictable data mining columns, and the Select training data step creates the input data mining columns.
Data Mining Column Structure
A data mining column is defined primarily by its data type and content type settings. These settings are detailed in other topics. Because of the diversity of possible data mining algorithm providers, the data mining column definitions are designed to be flexible and extensible.