DMSCHEMA_MINING_SERVICES Rowset

 

Applies To: SQL Server 2016 Preview

Provides a description of each data mining algorithm that the provider supports.

The DMSCHEMA_MINING_SERVICES rowset contains the following columns.

Column nameType indicatorDescription
SERVICE_NAMEDBTYPE_WSTRThe name of the algorithm. This column is provider-specific.
SERVICE_TYPE_IDDBTYPE_UI4This column contains a bitmap that describes the mining service. Microsoft SQL Server Analysis Services populates this column with one of the following values:

 DM_SERVICETYPE_CLASSIFICATION (1)

 DM_SERVICETYPE_CLUSTERING (2)
SERVICE_DISPLAY_NAMEDBTYPE_WSTRA localizable display name for the algorithm.
SERVICE_GUIDDBTYPE_GUIDThe GUID for the algorithm.
DESCRIPTIONDBTYPE_WSTRA user-friendly description of the algorithm.
PREDICTION_LIMITDBTYPE_UI4The maximum number of predictions the model and algorithm can provide.
SUPPORTED_DISTRIBUTION_FLAGSDBTYPE_WSTRA comma-delimited list of flags that describe the statistical distributions supported by the algorithm. This column contains one or more of the following values:

"NORMAL"

"LOG NORMAL"

"UNIFORM"
SUPPORTED_INPUT_CONTENT_TYPESDBTYPE_WSTRA comma-delimited list of flags that describe the input content types that are supported by the algorithm. This column contains one or more of the following values:

"KEY"

"DISCRETE"

"CONTINUOUS"

"DISCRETIZED"

"ORDERED"

"KEY SEQUENCE"

"CYCLICAL"

"PROBABILITY"

"VARIANCE"

"STDEV"

"SUPPORT"

"PROBABILITY VARIANCE"

"PROBABILITY STDEV"

"KEY TIME"
SUPPORTED_PREDICTION_CONTENT_TYPESDBTYPE_WSTRA comma-delimited list of flags that describe the prediction content types that are supported by the algorithm. This column contains one or more of the following values:

"KEY"

"DISCRETE"

"CONTINUOUS"

"DISCRETIZED"

"ORDERED"

"KEY SEQUENCE "

"CYCLICAL"

"PROBABILITY"

"VARIANCE"

"STDEV"

"SUPPORT"

"PROBABILITY VARIANCE"

"PROBABILITY STDEV"

"KEY TIME"
SUPPORTED_MODELING_FLAGSDBTYPE_WSTRA comma-delimited list of the modeling flags that are supported by the algorithm. This column contains one or more of the following values:

"MODEL_EXISTENCE_ONLY"

"REGRESSOR"

 

Note that provider-specific flags can also be defined.
SUPPORTED_SOURCE_QUERYDBTYPE_WSTRThis column is supported for backward compatibility.
TRAINING_COMPLEXITYDBTYPE_I4The length of time that training is expected to take:

 DM_TRAINING_COMPLEXITY_LOW indicates that the running time is relatively short, and it is proportional to input.

 DM_TRAINING_COMPLEXITY_MEDIUM indicates that the running time may be long, but it is generally proportional to input.

 DM_TRAINING_COMPLEXITY_HIGH indicates that the running time is long and it may grow exponentially in relationship to the number of training cases.
PREDICTION_COMPLEXITYDBTYPE_I4The length of time that prediction is expected to take:

 DM_PREDICTION_COMPLEXITY_LOW indicates that the running time is relatively short, and it is proportional to input.

 DM_PREDICTION_COMPLEXITY_MEDIUM indicates that the running time may be long, but it is generally proportional to input.

 DM_PREDICTION_COMPLEXITY_HIGH indicates that the running time is long and it may grow exponentially in relationship to the number of training cases.
EXPECTED_QUALITYDBTYPE_I4The expected quality of the model produced with this algorithm:

 DM_EXPECTED_QUALITY_LOW

 DM_EXPECTED_QUALITY_MEDIUM

 DM_EXPECTED_QUALITY_HIGH
SCALINGDBTYPE_I4The scalability of the algorithm:

 DM_SCALING_LOW

 DM_SCALING_MEDIUM

 DM_SCALING_HIGH
ALLOW_INCREMENTAL_INSERTDBTYPE_BOOLA Boolean that indicates whether the algorithm supports incremental training, i.e., updating the discovered patterns based on new factual data, rather than fully re-discovering the patterns.
ALLOW_PMML_INITIALIZATIONDBTYPE_BOOLA Boolean that indicates whether mining models can be created based on an PMML 2.1 string.

If TRUE, the mining algorithm supports initialization from PMML 2.1 content.
CONTROLDBTYPE_I4The support given by the service if training is interrupted:

 DM_CONTROL_NONE indicates that the algorithm cannot be canceled after it starts to train the model.

 DM_CONTROL_CANCEL indicates that the algorithm can be canceled after it starts to train the model, but must be restarted to resume training.

 DM_CONTROL_SUSPENDRESUME indicates that the algorithm can be canceled and resumed at any time, but results are not available until training is complete.

 DM_CONTROL_SUSPENDWITHRESULT indicates that the algorithm can be canceled and resumed at any time, and any incremental results can be obtained.
ALLOW_DUPLICATE_KEYDBTYPE_BOOLA Boolean that indicates whether cases can contain duplicate keys.

If VARIANT_TRUE, cases are allowed to contain duplicate keys.
VIEWER_TYPEDBTYPE_WSTRThe recommended viewer for this model.
HELP_FILEDBTYPE_WSTR(Optional) The name of the file that contains the documentation for this service.
HELP_CONTEXTDBTYPE_I4(Optional) The Help context ID for this service.
MSOLAP_SUPPORTS_ANALYSIS_SERVICES_DDLDBTYPE_WSTRThe version of DDL supported. 0 indicates no DDL support.
MSOLAP_SUPPORTS_OLAP_MINING_MODELSDBTYPE_BOOLA Boolean that indicates whether OLAP mining models can be created.

If TRUE, OLAP mining models can be created. Requires MSOLAP_SUPPORTS_ANALYSIS_SERVICES_DDL to be non-zero.
MSOLAP_SUPPORTS_DATA_MINING_DIMENSIONSDBTYPE_BOOLA Boolean that indicates whether data mining dimensions can be created.

If TRUE, data mining dimensions can be created.
MSOLAP_SUPPORTS_DRILLTHROUGHDBTYPE_BOOLA Boolean that indicates whether the service supports drillthrough capabilities.

If TRUE, the service supports drill-through capabilities.

The DMSCHEMA_MINING_SERVICES rowset can be restricted on the columns listed in the following table.

Column nameType indicatorRestriction State
SERVICE_NAMEDBTYPE_WSTROptional.
SERVICE_TYPE_IDDBTYPE_UI4Optional.

Data Mining Schema Rowsets

Community Additions

ADD
Show: