Page 177 - DCAP606_BUSINESS_INTELLIGENCE

Business Intelligence

Notes The customer identifier resident is the entity, and the purchases are the related attributes.

Dividing the Data into Training and Testing Sets

When you define the data for the mining structure, you can also specify that some of the data be
used for training, and some for testing. Therefore, it is no longer necessary to separate your data
in advance of creating a data mining structure. Instead, while you create your model, you can
specify that a certain percentage of the data be held out for testing, and the rest used for training.

Enabling Drillthrough

You can add columns to the mining structure even if you do not plan to use the column in a
specific mining model. This is helpful if, for example, you desire to get the e-mail locations of
customers in a clustering model, without using the e-mail address throughout the analysis
method.

Processing Mining Structures

When you process a mining structure, Analysis Services creates a cache that stores statistics
about the data, information about how any continuous attributes are discretized and other
information that is later used by mining models.

Viewing Mining Structures

In SQL Server Data Tools (SSDT), you can use the Mining Structure tab of Data Mining Designer
to view the structure columns and their definitions.

Self Assessment

State whether the following statements are true or false:

5. A single mining structure can support multiple mining models that share the identical
domain.
6. When you define a mining structure, you use columns that are available in an existing
data source view.
7. The building blocks of the mining structure are the mining structure columns, which
recount the data that the data source comprises.

8. A mining structure cannot comprise nested tables.

12.3 Configuring Algorithm Parameters

You can change the parameters supplied with the algorithms that you use to construct data
mining models to customize the results of the model. The algorithm parameters supplied in
Microsoft SQL Server Analysis Services change much more than just properties on the model,
they can be used to fundamentally adjust the way that data is processed, grouped, and displayed.

Example: You can use algorithm parameters to:
Change the procedure of analysis, such as the clustering procedure.
Control feature selection behaviour.

172 LOVELY PROFESSIONAL UNIVERSITY