ALL_MINING_MODEL_ATTRIBUTES

ALL_MINING_MODEL_ATTRIBUTES describes the attributes of the machine learning models accessible to the current user.

Only the attributes in the model signature are included in this view. The attributes in the model signature correspond to the columns in the training data that were used to build the model.

Machine learning models are schema objects created by Oracle Machine Learning for SQL.

Column Datatype NULL Description
OWNER VARCHAR2(128) NOT NULL Owner of the machine learning model
MODEL_NAME VARCHAR2(128) NOT NULL Name of the machine learning model
ATTRIBUTE_NAME VARCHAR2(128) NOT NULL

Name of the attribute

The target attribute name for ONNX models (mining model with ALGORITHM set to ONNX) is always ORA$ONNXTARGET.

ATTRIBUTE_TYPE VARCHAR2(12)  

Logical type of the attribute. The type is identified during the model build or apply process:

  • CATEGORICAL: Character data

  • MIXED: The input signature column takes on more than one attribute type. This is due to user-defined embedded transformations that allow an input column to be transformed into multiple independent mining attributes, including mining attributes of different types.

  • NUMERICAL: Numeric data

  • PARTITION: The input signature column is used for the partitioning key

  • TEXT: Unstructured text data

  • UNSTRUCTURED: Unstructured data

  • VECTOR: Attribute of type vectors (typically, target attribute of embedding models).
DATA_TYPE VARCHAR2(106)   Data type of the attribute
DATA_LENGTH NUMBER   Length of the data type
DATA_PRECISION NUMBER   Precision of a fixed point number. Precision, which is the total number of significant decimal digits, is represented as p in the data type NUMBER(p,s).
DATA_SCALE NUMBER   Scale of a fixed point number. Scale, which is the number of digits from the decimal to the least significant digit, is represented as s in the data type NUMBER(p,s).
USAGE_TYPE VARCHAR2(8)   Indicates whether the attribute was used to construct the model (ACTIVE) or not (INACTIVE). Some attributes may be eliminated by transformations or algorithmic processing. The *_MINING_MODEL_ATTRIBUTES view only lists the attributes used by the model, therefore the value of this column is always ACTIVE.
TARGET VARCHAR2(3)   Indicates whether the attribute is the target of a predictive model (YES) or not (NO). The target describes the result that is produced when the model is applied.
ATTRIBUTE_SPEC VARCHAR2(4000)  

One or more keywords that identify special treatment for the attribute during model build. Values are:

  • FORCE_IN: (GLM only) When feature selection is enabled, forces the inclusion of the attribute in the model build. Feature selection is disabled by default. If the model is not using GLM with feature selection enabled, this value is ignored.

  • NOPREP: When ADP is on, prevents automatic transformation of the attribute. If ADP is OFF, this value is ignored.

  • TEXT: Causes the attribute to be treated as unstructured text data. The TEXT value supports three subsettings: POLICY_NAME, MAX_FEATURES, TOKEN_TYPE, and MIN_DOCUMENTS. Subsettings are specified as name:value pairs within parentheses. For example: (POLICY_NAME:mypolicy)(MAX_FEATURES:2000)(TOKEN_TYPE:THEME). See Oracle Machine Learning for SQL API Guide for details.

  • NULL: The ATTRIBUTE_SPEC for this attribute is NULL.

    ATTRIBUTE_SPEC is a parameter to the PL/SQL procedure DBMS_DATA_MINING_TRANSFORM.SET_TRANSFORM. See Oracle AI Database PL/SQL Packages and Types Reference for details.

VECTOR_INFO VARCHAR2(56)  

Indicates the number of vectors and their data type in an ONNX model

Format: VECTOR(dimension,element_type)

Example: VECTOR(768,float32)