Data Analytics: Exploiting the Data Warehouse - Data Warehouse Systems: Design and Implementation - page 355

Node Probability Support Products.Name

... ... ... ...

72 0.0728 319 Assoc Line Items1(Raclette Courdavault)

73 0.0804 352 Assoc Line Items1(Gnocchi di nonna Alice)

74 0.0125 55 Assoc Line Items1(Sirop d ' erable)

74 0.0125 55 Assoc Line Items1(Sir Rodney ' s Scones)

75 0.0100 44 Assoc Line Items1(Zaanse koeken)

75 0.0100 44 Assoc Line Items1(Gnocchi di nonna Alice)

... ... ... ...

As another example, the following query returns the top 5 itemsets ordered by
the support for each node. The NODE CAPTION provides a text description
of the items:

SELECT TOP 5 (NODE SUPPORT), NODE NAME, NODE CAPTION
FROM Association.CONTENT
WHERE NODE TYPE = 7

This query returns the following table:

NODE SUPPORT NODE NAME NODE CAPTION

352 73 Gnocchi di nonna Alice = Existing

319 72 Raclette Courdavault = Existing

275 70 Gorgonzola Telino = Existing

275 71 Boston Crab Meat = Existing

264 69 Rhonbrau Klosterbier = Existing

Finally, we can use the model to issue prediction queries, typically based on
rules, to be used, for example, to make recommendations. The following query
tells us what products we can recommend to a customer who has purchased
a particular product. Since the column that corresponds to the products to
be predicted is a nested table, we must use one SELECT clause to map the
new value to the nested table column [Product Name] and another SELECT
clause to map the nested table column to the case-level column, [Assoc Line
Items] . Adding the keyword INCLUDE STATISTICS to the query displays the
probability and support for the recommendation:

SELECT FLATTENED PredictAssociation([Association].[Assoc Line Items],
INCLUDE STATISTICS, 3) AS [A]
FROM [Association] NATURAL PREDICTION JOIN
( SELECT
( SELECT ' Tarte au sucre ' AS [Product Name] )
AS[AssocLineItems])AST