Oracle 视图 ALL_MINING_MODEL_ATTRIBUTES 官方解释,作用,如何使用详细说明
  iDU31ygkXmx7 2023年11月21日 14 0

本站中文解释

ALL_MINING_MODEL_ATTRIBUTES视图是Oracle 数据挖掘中用于查看模型属性信息的视图,包括训练出来的模型的描述、模型技术和模型参数等。

以查看算法训练出的模型的属性信息为例,使用该视图的步骤如下:

1. 以用户名USER_NAME、模型名MODEL_NAME为查询条件,调出ALL_MINING_MODEL_ATTRIBUTES视图,可看到模型的各种属性信息,包括名称、类型、算法、参数等信息。

2. 调出模型的参数信息LEARN_PARAMETERS。

3. 调出模型的准确度信息ACCURACY。

以上这些就是视图ALL_MINING_MODEL_ATTRIBUTES 所能提供的有关模型的属性信息,可以用来进行算法模型评价和挖掘结果回放、可视化等操作。

官方英文解释

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.

Related Views

  • DBA_MINING_MODEL_ATTRIBUTES describes the attributes of all machine learning models in the database.

  • USER_MINING_MODEL_ATTRIBUTES describes the attributes of the machine learning models owned by the current user. This view does not display the OWNER column.

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

ATTRIBUTE_TYPE

VARCHAR2(11)

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

  • NUMERICAL: Numeric data

  • CATEGORICAL: Character data

  • TEXT: Unstructured text data

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

  • 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.

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 Database PL/SQL
    Packages and Types Reference
    for details.

See Also:

Oracle Machine Learning
for SQL API Guide
for more information about the attributes of machine learning models

【版权声明】本文内容来自摩杜云社区用户原创、第三方投稿、转载,内容版权归原作者所有。本网站的目的在于传递更多信息,不拥有版权,亦不承担相应法律责任。如果您发现本社区中有涉嫌抄袭的内容,欢迎发送邮件进行举报,并提供相关证据,一经查实,本社区将立刻删除涉嫌侵权内容,举报邮箱: cloudbbs@moduyun.com

  1. 分享:
最后一次编辑于 2023年11月21日 0

暂无评论

iDU31ygkXmx7