Data mining — Mining data specification

You must select and specify the data that you want to use for building or testing mining models. The specification of the input data includes the information that is required for the training run, for example, the name of the source table and information about the columns in this table.

Specifying mining data is the first step in the model-building process. In this step, the following tasks are covered:

Creating values of the data type DM_MiningData, perform operations on them, or both.
Using the user-defined method DM_defMiningData.

The data type DM_MiningData stores the metadata to access the physical data in source tables. These source tables can be either base tables or views. The metadata consists of the fully-qualified name of the source table, the column names, and the SQL data types of the source data.

In addition, DM_MiningData stores any other information necessary to make the physical data accessible for a subsequent training or test run. This other information might include alias names or, optionally, a WHERE clause to filter the data. The alias name maps a field name in the physical data to a field name in a logical data specification. An alias name and its corresponding field name in the logical data specification must be the same. The default arrangement is that the alias names are the same as the names in the physical data. Use alias names that are different from the names in the physical data only if a logical data specification with different names already exists.