Validating Mining Models

Although some of our clients have validated the usefulness of their data mining models simply by using the viewers available on the Mining Model Viewer tab to help them understand the results of the mining model algorithm, most consumers of data mining projects also want to include an accuracy check of the models that are produced. This is done to validate the output produced and to guide further customization of a particular model—or even to help determine which algorithms shed the most light on the particular business situations being addressed. You should be aware that in the CRISP-DM software development life cycle, there is a distinct phase of the software development life cycle dedicated to model evaluation because ...

Get Smart Business Intelligence Solutions with Microsoft® SQL Server® 2008 now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.