Package oracle.dmt.odm.task

This package contains Java classes representing mining tasks.

See:
Description

Interface Summary

ExecutionHandle	An ExecutionHandle serves as a handle for the application that can be used to monitor or cancel a task execution.
TaskStatusHistory	The interface TaskStatusHistory is an ordered list iterator of task execution statuses.

Class Summary

ClassificationTestTask	An instance of ClassificationTestTask is used for testing a model on test data.
CrossValidateTask	An instance of CrossValidateTask provides an additional technique for measuring the accuracy of a predictive model.
MiningApplyTask	An instance of MiningApplyTask is used for applying a model to a data set to make predictions, classifications, and to provide associated probabilities.
MiningBuildTask	An instance of MiningBuildTask is used for building all mining models supported by ODM.
MiningDataTask	The abstract class MiningDataTask is the common superclass for all mining tasks that involve data mining operations e.g, (building models, scoring(apply), testing, cross-validate, lift computation).
MiningLiftTask	An instance of MiningLiftTask is used to compute the lift based on the specified positive target value and the number of quantiles.
MiningTask	The abstract class MiningTask is the common superclass for all data mining task classes.
MiningTaskStatus	An instance of MiningTaskStatus provides the following details on the state of a tasks execution: MiningTaskState enumeration State entry timestamp State description A given task may have multiple MiningTaskStatus instances that provide a status history for the task .
ModelExportTask	An instance of ModelExportTask is used to export ODM mining models to standard format mining model representations.
ModelImportTask	An instance of ModelImportTask is used to import standard format mining models to the database.
ModelSeekerTask	Deprecated. Model Seeker functionality is deprecated in version 10.1.
RegressionTestTask	An instance of RegressionTestTask is used for testing an Regression (regression) model on test data.

Exception Summary

MiningTaskException

The exception MiningTaskException is thrown when there is a failure during the execution of a mining task.

Package oracle.dmt.odm.task Description

This package contains Java classes representing mining tasks. Mining tasks are named objects that perform a designated mining operation asynchronously. A task object serves as a container within which to specify arguments to data mining operations to be performed by Oracle Data Mining. By providing an object to specify tasks, deferred (batch) processing of tasks is feasible. The task execution handle can be used to interrupt or terminate executing invocations. For deferred or batch processing, one or more task objects can be saved and scheduled for execution at a later time.

Note that each type of mining model also provides synchronous methods for appropriate mining operations. For example, ClusteringModel provides static methods for build and apply. Mining model classes also provides other static methods such as apply, test, and compute lift (for supervised and clustering models), and for association rules model, various rule retrieval methods.

ODM supports the following mining operations. Refer to the Concepts manual for more details.

• Model building: A MiningModel is built in the data mining server (DMS). After the model is built, it is persisted in the DMS and can be accessed by its user specified unique name. Costs and priors may be provided when building a classification model. (Refer to MiningBuildTask.)
• Test: Model testing gives an estimate of model accuracy. Only classification models may be tested in ODM. The test results are stored in a mining test result objects.
• Apply (scoring): Applying a supervised learning model (i.e., classification models in ODM) to data produces scores or predictions with an associaed probability. Applying a ClusteringModel to data results in, for each record, a predicted cluster ID and the probability that the record is in that cluster. The data to be scored must have attributes compatible with those of the training data; it must have the same attributes with the same names and respective data types or a suitable mapping. The output of apply operation is placed in the schema specified by the user. (Refer to MiningApplyOutput and MiningApplyTask.)
• Lift: Given a designated positive and negative values, test cases are sorted according to how confidently they are predicted to be positive instances (in descending order of confidence). Based on that ordering, they are partitioned into a specified number of quantiles. Then, a various set of statistics is computed.

ODM supports the following types of mining task:

• MiningBuildTask: performs model build operation with parameters including an input data, a MiningFunctionSettings, and the name of the model
• MiningApplyTask: performs apply operation with parameters including an input data, a model name to be applied, an apply output specification and its location, and the name of the apply result metadata.
• ClassificationTestTask: performs classification test mining operation with parameters including an input data, a model name to be tested, and the name of the test result metadata.
• ApproximationTestTask: performs approximation test mining operation with parameters including an input data, a model name to be tested, and the name of the test result metadata.
• MiningLiftTask: performs lift mining operation with parameters including an input data, the number of quantiles, the positive target values, a model name to be applied, and the name of the lift result metadata.
• CrossValidateTask: performs cross validation mining operation with parameters including an input data, a function settings name, and the name of the result metadata. This task effectively builds a NaiveBayesModel and runs cross validation against the model to measure the accuracy of the model (using the same input data).
• ModelSeekerTask: used to build a set of models and performs test and/or lift mining operations on them to find a better model.
• ModelExportTask: used to export a model as a PMML v2.0 string and persist the string in a specified location. Oracle Data Mining supports exporting association rules models and naive bayes models into PMML v2.0.
• ModelImportTask: used to import a model of a specified type and persist in the DMS with a designated name. Oracle Data Mining supports importing association rules models and naive bayes models in PMML v2.0.