The data used to calculate the test metrics for a classification or regression model must reside in an Oracle 10g table or view. The data must be compatible with the data used to build the model. It should contain values for the target attribute. The metrics measure how well the predicted values for the target match known values.
Oracle Data Miner supports the following Oracle datatypes:
NUMBER
VARCHAR2
CHAR
DATE
CLOB
, BLOB
, BFILE
, LONG
, VARCHAR2
, XMLType
, CHAR
, RAW
, or LONG RAW
for text data only. For information about text mining, see Text Mining. Attributes of type DATE
are automatically converted to numbers before mining is performed by calculating the Julian date. For details, see Date Conversion.
Any column in a table that has a datatype not in this list is ignored by Oracle Data Miner. The columns are automatically filtered out and are, therefore, not displayed.
Specify the location of the data used to calculate the metrics:
Click Next to continue. The wizard performs the following validations:
If the data for calculating is not compatible with the data used to build the model, the differences are reported.
The wizard then extracts the attributes.
Copyright © 2006, 2008, Oracle. All rights reserved.