Predicts associative membership.


This is different from creating a prediction on an association model.


PredictAssociation(<table column reference>, option1, option2, n ...)

Applies To

Classification algorithms and clustering algorithms that contain predictable nested tables. Classification algorithms include the Microsoft Decision Trees, Microsoft Naive Bayes, and Microsoft Neural Network algorithms.

Return Type

<table expression>


The options for the PredictAssociation function include EXCLUDE_NULL, INCLUDE_NULL, INCLUSIVE, EXCLUSIVE (default), INPUT_ONLY, INCLUDE_STATISTICS, and INCLUDE_NODE_ID.


INCLUSIVE, EXCLUSIVE, INPUT_ONLY, and INCLUDE_STATISTICS apply only for a table column reference, and EXCLUDE_NULL and INCLUDE_NULL apply only for a scalar column reference.

INCLUDE_STATISTICS only returns $Probability and $AdjustedProbability.

If the numeric parameter n is specified, the PredictAssociation function returns the top n most likely values based on the probability:

PredictAssociation(colref, [$AdjustedProbability], n)

If you include $AdjustedProbability, the statement returns the top n values based on the $AdjustedProbability.


The following example uses the PredictAssociation function to return the four products in the Adventure Works database that are most likely to be sold together.

  PredictAssociation([Association].[v Assoc Seq Line Items],4)

See Also


Data Mining Extensions (DMX) Function Reference
Functions (DMX)
Mapping Functions to Query Types (DMX)

Help and Information

Getting SQL Server 2005 Assistance