Creating DMX Prediction Queries
The main goal of most data mining projects is to use mining models to create predictions for new data. For example, you may want to predict how many bicycles your company will sell next year during December, or whether a potential customer will purchase a bicycle in response to an advertising campaign. You can also use predictions to explore the information that the algorithms discover when they train the mining models.
Prediction queries are based on the Data Mining Extensions (DMX) language. DMX extends the SQL language, to provide support for working with mining models.
SQL Server provides two tools that you can use to build prediction queries: Prediction Query Builder and the Query Editor. Prediction Query Builder is included in the Mining Model Prediction tab of Data Mining Designer. When you use the query builder, you can use graphical tools to design a query, use a text editor to manually modify the query, and use a simple results pane to view the results of the query. The Query Editor provides tools that you can use to build and run DMX queries.
You can also include prediction queries as part of a SQL Server 2005 Integration Services (SSIS) package.