Data Mining Query Task

boneyardtherapeuticData Management

Nov 20, 2013 (4 years and 6 months ago)




June 23, 2013

Data Mining Query Task

The Data Mining Query task runs prediction queries based on data mining models built in
Analysis Services. The prediction query creates a prediction for new data by using mining
models. For example, a prediction query can predict
how many sailboats are likely to sell during
the summer months or generate a list of prospective customers who are likely to buy a sailboat.

Integration Services provides tasks that perform other business intelligence operations, such as
running Data Defin
ition Language (DDL) statements and processing analytic objects.

Prediction Queries
The query is a Data Mining Extensions (DMX) statement. The DMX language is an
extension of the SQL language that provides support for working with mining models.

The task
can query multiple mining models that are built on the same mining structure. A mining
model is built using one of the data mining algorithms that Analysis Services provides. The
mining structure that the Data Mining Query task references can include multi
ple mining models,
built using different algorithms.

The prediction query that the Data Mining Query task runs returns a result that is a single row or
a data set. A query that returns a single row is called a singleton query: for example, the query
that p
redicts how many sailboats will be sold during the summer months returns a number.

The query results are saved to tables. If a table with the name that the Data Mining Query task
specifies already exists, the task can create a new table, using the same na
me with a number
appended, or it can overwrite the table content.

If the results include nesting, the result is flattened before it is saved. Flattening a result changes
a nested result set to a table. For example, flattening a nested result with a

column and
a nested

column adds rows to the

column to make a table that includes
product data for each customer. For example, a customer with three different products becomes a
table with three rows, repeating the customer in each row a
nd including a different product in
each row. If the FLATTENED keyword is omitted, the table contains only the

and only one row per customer.

Configuration of the Data Mining Query Task


The Data Mining Query task requires two connections.

The first connection is an
Analysis Services connection manager that connects either to an instance of Microsoft


Analysis Services or to an Analysis Services project that contains the mining structure
and the mining model. The second connection

is an OLE DB connection manager that connects
to the SQL Server database that contains the table to which the task writes.

You can set properties through SSIS Designer or programmatically.