Results for “odm_oc_clustering_model”

50+ results




AI-generated from documented ETRM metadata — verify critical details on the linked pages.

Overview

ODM.ODM_OC_CLUSTERING_MODEL is a wrapped PL/SQL package belonging to the Oracle Data Mining (ODM) schema, the database-resident mining engine embedded in Oracle E-Business Suite and the Oracle Database. In the EBS 12.1.1 and 12.2.2 environments, this package implements the One-Class Clustering (OC) model build path of the ODM PL/SQL API, the PL/SQL counterpart to the Java-based ODM clients and to the DBMS_DATA_MINING package. Its business function is to construct a clustering model that describes a "normal" population and identifies outliers or anomalies relative to that population — an approach that is used in EBS for exception detection, fraud screening, supplier and expense anomaly detection, and quality profiling, rather than for the more conventional multi-cluster segmentation.

The package is classified as OTHER in the ETRM 12.2.2 metadata and is delivered wrapped, meaning the source is obfuscated and only the public specification is visible. The extract confirms this: the body shows the wrapped marker and repeated obfuscated blocks, while the specification lists the callable interface.

Key Procedures and Functions

ETRM documents zero procedures or functions for this object, so no parameter lists are asserted here. However, the wrapped specification embedded in the extract exposes the following named entry points and constants:

Tables Accessed

The ETRM metadata lists no APPS-synonym tables referenced by this package. From the documented parameters, the operative data objects are:

  • A training table supplied by the caller, read through the TID, item and value column mapping — typically a transactional or attribute-value table in the mining schema.
  • A bins/attribute type table (P_BINS_ATTR_TYPE_TABLE, P_BINS_TABLE) describing discretisation of numeric attributes.
  • A cluster output table (P_CLUSTER_TABLE_NAME) created by CREATE_PERSIST_TABLE, holding the persistent cluster nodes, leaves and tree levels.
  • ODM repository tables recording the mining task (P_MINING_TASK_ID via MINING_TASK_ID_TYPE) and, as the extract indicates, an internal dependency on ODM_MODEL_UTIL for shared model utilities.

Usage Notes

ODM_OC_CLUSTERING_MODEL is not exposed through an EBS form; it is invoked programmatically. Typical call sites are ODM's own PL/SQL mining API layers, the Applications Technology data-mining stack, and custom PL/SQL that builds anomaly-detection models against EBS data. It may be driven from a concurrent program or a scheduled job that calls BUILD_OCLUSTER to refresh the model, followed by scoring calls to OCLUSTER. The ETRM metadata records that it is referenced by one other package, consistent with its role as a lower-level engine invoked by a higher-level ODM driver. Because the package is wrapped and ODM is a licensed database option, its use requires the Data Mining option to be installed, and the model build must run under a database account with execute privilege on the ODM schema objects and appropriate quota on the mining tablespace.