Vendor Implementations - Java Data Mining: Strategy, Standard, and Practice - page 413

Table 16-3 Oracle Java Data Mining extensions

OJDM extension class/package Description

oracle.dmt.jdm.resource
OraConnectionFactory

This class provides an easy way to create Oracle-specific
DME ConnectionFactory objects. It is particularly useful
when applications do not have JNDI access.

oracle.dmt.jdm.featureextraction
OraFeatureExtractionSettings
OraFeatureExtractionModel
OraFeature
OraFeatureExtractionApplySettings
oracle.dmt.jdm.algorithm.nmf
OraNMFAlgorithmSettings

These classes and packages are used to define the feature
extraction function and the Non-Negative Matrix
Factorization (NMF) algorithm. OraFeatureExtraction-
Settings is the build settings used to specify the maximum
number of features that should be derived by the model.
OraNMFAlgorithmSettings specifies NMF algorithm
specific settings. OraFeatureExtractionModel and
OraFeature classes represent the model that defines the
extracted features. OraFeatureExtractionApply
Settings is used to define the apply settings for feature
extraction.

oracle.dmt.jdm.algorithm.abn
OraABNSettings
oracle.dmt.jdm.modeldetails.abn
OraABNModelDetail

These classes and packages are used to define Oracle
proprietary Adaptive Bayes Network (ABN) algorithm
settings and associated model details.

oracle.dmt.jdm.algorithm.ocluster
OraOClusterSettings

This package and class are used to define Orthogonal
Partitioning Cluster (O-Cluster) algorithm settings. The
JDM standard defined generic clustering model is sufficient
to represent O-Cluster model.

oracle.dmt.jdm.task
OraPredictTask
OraExplainTask

OJDM offers these non-JDM tasks that automate data
mining processes in the database and produce final
output. OraPredictTask is used to predict the target
attribute of the input table/view. This task automates the
model build, test, and apply operations and produces the
final predictions as an output table. OraExplainTask is
used to explain the specified target attribute of the input
table/view. This task produces an output table that
contains the attribute importance and rank. These two tasks
produce reasonable quality results with minimum effort.

oracle.dmt.jdm.transform
oracle.dmt.jdm.transform.binning
oracle.dmt.jdm.transform.clipping
oracle.dmt.jdm.transform.normalize
oracle.dmt.jdm.transform.text

These packages are used to define the transformation-
related classes. OJDM defines some of the common data
mining data transformations, such as binning/discretiza-
tion, clipping/outlier handling, normalization. Text data
transformations convert text data in the datatypes such as
CLOB to a nested table structure that can be mined using
OJDM.