Data Preparation

Mining Build Activities include steps that perform data preparation required by the algorithm that you selected. It is always possible to perform required transformations "by hand" without using the activity. Oracle Data Miner provides transformations to perform required transformations as well as other transformation, such as Recode, that may be required by mining problems.

The Data menu provides a collection of wizards that perform common data preparation steps. SQL Worksheet on the Tools menu may also be useful during data preparation.

The Data menu contains the selections Copy Table, Create Table From View, Create View (a visual query builder), Generate SQL, Import From File, Show Lineage, Show Summary Single-Record, Show Summary Multi-Record, the Transform submenu, Predict, and Explain.

Note: Predict and Explain implement predictive analytics; see Predictive Analytics for details.

Transform Submenu

The Transform submenu of the Data menu contains the following transformations: Aggregate, Compute Field, Discretize (bin), Filter Single-Record case table, Missing Values, Normalize, Numeric, Outlier Treatment, Recode, Sample, Stratified Sample, Split, Stratified Split, Variation Filter, and Text.

Invoking Transforms

Many of the selections in the Data menu can be invoked from the context menu that appears when you select a table or view and right click. In this case, the selected table or view is automatically set as the input of the transformation.