Results for “odm_km_clustering_model”

50+ results




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

Overview

The ODM.ODM_KM_CLUSTERING_MODEL package is a wrapped PL/SQL package that belongs to the Oracle Data Mining (ODM) schema shipped with Oracle E-Business Suite 12.1.1 and 12.2.2. Its business function is to build unsupervised clustering models against mining data stored in the EBS database. In practical terms, it implements the server-side engine behind Oracle Data Mining's clustering algorithms: k-Means, k-Modes, k-Hybrid, O-Cluster (orthogonal partitioning), density-based clustering, and c-Means. Within the Oracle Advanced Analytics / Data Mining stack embedded in EBS, this package transforms a flat training table of attributes into a hierarchical clustering structure, producing cluster identifiers, leaf assignments, and cluster statistics that downstream applications use for segmentation, classification, anomaly detection, and customer grouping.

The package is classified as OTHER in the ETRM metadata, reflecting that it is an internal engine rather than a public integration API. The "wrapped" designation indicates the source is obfuscated by the Oracle PL/SQL wrapper, so only the specification boundary and internal identifier names are externally visible.

Key Procedures and Functions

The ETRM metadata documents zero exposed procedures or functions for this package, and the wrapped body exposes only internal identifiers rather than a callable public specification. The decodable internal names do reveal the functional design:

Internal collection types (CLUSTER_NODE_TYPE, CLUSTER_INTERNAL_TYPE, HISTOGRAM_TAB_TYPE, CELL_TYPE, BINS_TYPE) and the ATTRIBUTE_TYPE record (SEQUENCE_ID, ATTRIBUTE_ID, VALUE, ATTRIBUTE_NAME, NUMBINS) form the in-memory model built during training. The constants C_MAX_POINTS (2000) and C_EPSILON (0.0001) bound memory use and convergence tolerance.

Tables Accessed

No application tables are documented in the ETRM metadata for this package. The visible direct references are to DBMS_UTILITY (used for call-stack and error diagnostics, consistent with the trace routine) and PLITBLM (the PL/SQL implementation of DBMS_UTILITY table helpers). The package additionally accepts P_TRAINING_TABLE_NAME and P_CLUSTER_TABLE_NAME as dynamic table-name parameters, delegating physical reads and writes to the mining tables named at runtime—typically the ODM mining data and model output tables managed by ODM_MODEL_UTIL, which is referenced in the signature.

Usage Notes

This package is not intended for direct invocation from EBS forms, concurrent programs, or custom code. It is an internal engine invoked by the Oracle Data Mining model-building layer, which constructs the training table, selects the algorithm, and calls the build routine. The P_MINING_TASK_ID parameter ties each build to a mining task record, and the logging hooks exist to support ODM's debugging and performance tracing. Because the body is wrapped and the ETRM classification is OTHER, the package should be treated as closed to customization. The reference to an identity/account/login URL in the originating search is unrelated to this database object and does not reflect any documented behavior of ODM_KM_CLUSTERING_MODEL.