| title | description | author | ms.author | ms.reviewer | ms.date | ms.service | ms.subservice | ms.topic | ms.custom |
|---|---|---|---|---|---|---|---|---|---|
Modeling Flags (DMX) |
Modeling Flags (DMX) |
minewiskan |
owend |
owend |
02/17/2022 |
sql |
analysis-services |
reference |
dmx |
[!INCLUDEssas]
You can use modeling flags in [!INCLUDEssASnoversion] to provide additional information to a data mining algorithm about the data that is defined in a case table. The algorithm can use this information to build a more accurate data mining model. You can define modeling flags on both mining structure columns and mining model columns.
[!INCLUDEssASnoversion] supports the following modeling flags:
NOT NULL
The values for the attribute column should never contain a null value. An error will result if [!INCLUDEssASnoversion] encounters a null value for this attribute column during the model training process. This flag is defined on a mining structure column.
REGRESSOR
Indicates that the algorithm can use the specified column in the regression formula of regression algorithms. This flag is supported by the [!INCLUDEmsCoName] Linear Regression and [!INCLUDEmsCoName] Decision Trees algorithms, and is defined on a mining model column.
MODEL_EXISTENCE_ONLY
The values for the attribute column are less important than the presence of the attribute. This flag is defined on a mining model column.
Third-party algorithms may support additional modeling flags. To determine which modeling flags an algorithm supports, use the SUPPORTED_MODELING_FLAGS schema rowset. You can also query the mining services on the server to determine which modeling flags are supported for a particular algorithm. For example, the following query returns the modeling flags are supported for the Microsoft Linear Regression algorithm on the current server:
SELECT SUPPORTED_MODELING_FLAGS
FROM $SYSTEM.DMSCHEMA_MINING_SERVICES
WHERE SERVICE_NAME = 'Microsoft_Linear_Regression'
Expected results:
NOT NULL,REGRESSOR
For examples of the syntax that [!INCLUDEssASnoversion] supports for specifying a flag on a mining structure column, see CREATE MINING STRUCTURE (DMX).
For an example of the syntax for specifying a modeling flga on a mining model column, see ALTER MINING STRUCTURE (DMX).
For more information about working with mining model columns, see Mining Model Columns.
Data Mining Algorithms (Analysis Services - Data Mining)
Data Mining Extensions (DMX) Reference
Data Mining Extensions (DMX) Syntax Elements
Data Mining Extensions (DMX) Function Reference
Data Mining Extensions (DMX) Operator Reference
Data Mining Extensions (DMX) Statement Reference
Data Mining Extensions (DMX) Syntax Conventions
General Prediction Functions (DMX)
Structure and Usage of DMX Prediction Queries
Understanding the DMX Select Statement