After you create a model, you can create a test task to verify the accuracy of the model. This topic describes the process of creating a test task.
Data mining models are built using known data to train the algorithm so it can be applied to a similar data set. To test a model, you create a test task. In the test task, you specify a model you have trained and a set of accessor domain values. In addition to the predictor and target accessors, you specify test accessors that reference a known set of results.
The task of building a test model includes the following steps:
Select the database to use.
Select the model to apply.
Select the set of data and parameters to test the algorithm.
Execute the algorithm against the data to generate test results.
The test task compares the results of the trained model to the set of known results you specify. The test task determines if the results match within the range specified by the test parameters. If the results do not match, you can do any of the following:
To build or modify a test model using Mining Wizard:
Select File, then Wizards, and then Data Mining Wizard to launch the Mining Wizard.
Select the server, the application, and the database to use and click Next.
If Advanced mode is desired, select Advanced, then complete one of the following actions:
Select Create a New Test Task, click Next, and go to Step 4.
Select Modify a Saved Test Task, click Next, and go to Step 5.
If you are creating a task, from the list of saved build task models, select a model to use and click Next.
The Algorithm column shows the name of the algorithm that was used to build the model.
If you are modifying a saved task, from the list of saved test task templates, select a test task and click Next.
In the left pane, expand all nodes, and then complete the following steps:
If you want to change the default setting for how the algorithm handles missing values, select Choose mining task setting, and, in the right-hand pane, change the setting.
See Enter algorithm parameter and accessor information for detailed information.
Select Specify parameter values for the task to set task parameters.
Test tasks have one or both of the following parameters: Deviation specifies the amount that the test results can deviate from the known results and still be considered accurate. This parameter requires a double value and only applies to algorithms that return numerical, not categorical, results.Confidence specifies the ratio of individual test results that must match the known data, compared to the total number of tests, for the test task to validate the results generated by the algorithm. The parameter requires a double value in the range of 0-1, where 1 specifies all test results must match the known data. All algorithms require this parameter.
Select each attribute in turn to view the MDX expression, if any, entered for the attribute. The values from the build model are automatically provided for the apply model.
Provide values for the following two attributes to limit the range of test data to consider: Test.External defines the scope of the test data. Test.Anchor specifies additional restrictions from other dimensions.
Click Next.
Perform an action:
Select Save and execute mining task.
Select Save mining task.
Select Execute mining task.
If you select Save mining task or Save and execute mining task, enter the Template name. The template is stored in the Data Mining—Mining Templates node. Templates enable you to reuse and modify information to build a series of models. You can execute saved models from the Mining Templates node. See Managing Data Mining Templates for information.
If you select Execute mining task or Save and execute mining task, enter the Mining Result and, if desired, a Mining Result Comment.
Click Finish to execute the task and/or save the task template.
To view the progress of the test task execution, see Managing Data Mining Sessions for details.
To view test task results, see Managing and Viewing Data Mining Results for details.
“MDX” in the Oracle Essbase Technical Reference
Mining an Essbase Database in the Oracle Essbase Database Administrator's Guide
create mining task template (MaxL)
display mining session (MaxL)
alter system stop mining session (MaxL)