Summary
The technology known as data warehousing has developed as a means of making use of the huge quantities of data available these days. In addition to storing all this data, data mining is needed to make useful predictions about it. Analysis Services, a separate install for SQL Server 2000, offers data-mining capabilities in an easy-to-use and scalable way.
This chapter is one of two that examines the effort of a fictional discount retailer named Savings Mart to improve its operational efficiencies. A sample application named LoadSampleData, provided on the book’s Web site, allows readers to generate a unique dataset for the data-mining model. Optionally, the reader can also attach a database file provided on the Web site.
One of the biggest problems affecting successful data mining is invalid or incorrect data. Therefore, the process of cleaning a database is often the most time-consuming aspect of preparing a dataset.
We step through the process of creating a mining model using Analysis Services. This involves creating a database, naming the data source, and using the mining-model wizard to create the actual model.
Once a model is created, it can be trained with a training dataset to produce prediction results. The training dataset in this chapter represents one year’s worth of purchases and shipments to all five stores. These results will be the basis for a Windows service created in the next chapter.