OLE DB for Data Mining Compliance

Analysis Services

previous page next page
Analysis Services

OLE DB for Data Mining Compliance

This section contains an overview of the differences between the features of this version of Microsoft® SQL Server™ 2000 Analysis Services and the OLE DB for Data Mining specification of June 2000.

Analysis Services complies with all but eight items required by the specification.

Analysis Services extends the MINING_MODELS schema rowset with an additional column.

Noncompliant Items

Of the eight items that do not comply with the OLE DB for Data Mining specification, four deal directly with supported data mining grammar, two refer to supported data mining model settings, and two refer to data mining functions.

Item Analysis Services implementation OLE DB for Data Mining specification
Refining Mining Models All of the data mining model content must be deleted and the data mining model must be retrained using the full set of old and new cases. A data mining model can be refined by executing another INSERT INTO statement with additional cases, if the mining algorithm provider supports refining data mining models.
CONTENT_TYPE CONTENT_TYPE supports the following content types:
  • CONTINUOUS

  • DISCRETE

  • DISCRETIZED

  • KEY

  • TABLE
 CONTENT_TYPE supports the following content types:
  • CONTINUOUS

  • CYCLICAL

  • DISCRETE

  • DISCRETIZED

  • KEY

  • ORDER

  • ORDERED

  • PROBABILITY

  • PROBABILITY_VARIANCE

  • PROBABILITY_STDEV

  • SEQUENCE

  • SEQUENCE_TIME

  • STDEV

  • SUPPORT

  • TABLE

  • VARIANCE
DISTRIBUTION_FLAG DISTRIBUTION_FLAG supports the following distribution types:
  • NORMAL
 DISTRIBUTION_FLAG supports the following distribution types:
  • BINOMIAL

  • LOG_NORMAL

  • MULTINOMIAL

  • NORMAL

  • POISSON

  • T-DISTRIBUTION

  • UNIFORM
INSERT Statement <source data query> supports the following source data query types:
  • OPENROWSET

  • SHAPE

  • SINGLETON SELECT
 <source data query> supports the following source data query types:
  • OPENROWSET

  • SELECT

  • SHAPE

  • SINGLETON CONSTANT

  • SINGLETON SELECT
SELECT Statement SELECT * FROM <model> syntax not allowed. SELECT * FROM <model> syntax used to browse all possible cases.
CREATE MINING MODEL Statement The following modeling flags are supported in the column definition statement for handling missing values:

NOT NULL

 The following modeling flags are supported in the column definition statement for handling missing values:

NOT NULL
IGNORE NULL
NULL INFORMATIVE
PredictScore(<scalar column reference>) Not implemented. For predictable columns, this function returns the prediction score, as a scalar value, of the input dataset on the specified column.
PredictNodeID(<scalar column reference>) Not implemented. For predictable columns, this function returns the node ID of the leaf node where the case is classified for the specified column.
PredictProbabilityStdev(<scalar column reference>) Not implemented. This function returns the standard deviation of probability for the histogram entry with the highest probability.
PredictProbabilityVariance(<scalar column reference>) Not implemented. This function returns the variance of probability for the histogram entry with the highest probability.

Additional Items and Extensions

Analysis Services provides additional functionality that extends the OLE DB for Data Mining specification by implementing an additional schema rowset column.

Schema Rowset Extensions

The MINING_MODELS schema rowset stores meta data related to data mining models. The following table contains columns added to the MINING_MODELS schema rowset.

Column name Type indicator Description
MSOLAP_MODEL_SOURCE DBTYPE_WSTR Stores the name of the source cube for OLAP mining models.

previous page start next page