The SQL Server 2005 Data Mining Addins for Office 2007 includes a delightful model rendering template for Visio 2007.  With this template, you can render your graphical views from SQL Server Data Mining as Visio diagrams for annotation, enhancement, and presentation.
The Data Mining Template for Visio 2007 allows you to render decision trees, regression trees, cluster diagrams, and dependency nets.  While most of the behavior of the built-in viewers are preserved in the Visio environment, you also get additional functionality such as the ability to move sub-trees to new pages, grow dependency networks starting at an arbitrary node, or create a cluster diagram using discrimination charts.
Once rendered, you can use the power of Visio to annotate your diagrams with text, add supplemental graphics supporting your model, and even apply color themes that match your presentation needs.  Finally the models can be presented as Visio documents, embedded into other Office documents or simply saved as a web page.

The links here are to models saved as web pages as examples of some of the possibilities.  New links will be added as they are created.

Home Ownership Decision Tree

Customer Clusters

Movies Dependency Network

To start the sample click here

This sample shows how to cluster in-memory data using the session mining models feature. Using the form you can layout the size, shape, and density of up to 6 ellipses - or you can try out one of the predefined layouts. You can also specify up to 0.8% random noise in the data by adjusting the Noise Level parameter. To gain a better understanding of how the clustering algorithm works, you can tune all the clustering parameters to see how they affect the results. When ready, click the Find Clusters button to see the results. When this button is clicked, the page generates an in-memory data set on the server according to the layout you specified. It then creates a session mining model on the server and trains it by sending the in-memory data using a rowset parameter. The result is displayed on the page by coloring each cluster a unique color. You can choose three different ways the clusters are shaded. The first shading method is Areas, which colors data points according their cluster probability, taking into account the likelihood of each individual cases – areas where cases do not exist will appear white. The second approach, Soft Regions, works similar to Areas, but it ignores case likelihood. This creates a display that is fully colored with the colors blended where clusters overlap. The last way, Hard Regions, simply colors the entire region using the color of the most probable cluster.

The source code for this application is available here.

You may also want to check out two useful tips on mining in-memory data and working with temporary models.

To launch the sample, click here.

This sample allows building DMX (data mining prediction) statements on the fly. The statements are encapsulated in XMLA (XML For Analysis) packages and sent to the Analysis Services server. The server response is also an XMLA package. The sample application parses the response XMLA and displays the result in a table.

To run the sample:

• Click here to launch the sample in a separate window
• Select a database in the left list
• Select a model in the right list
• Click on the Load XMLA Page button
  The Based On column contains the input columns of the mining model with their respective values Select or enter a value for the columns you want to use as input
  The Predict column contains the predictable columns of the mining model. Select one
• Check the "Show DMX" check box to ensure that the generated query will be displayed
• Check the "Show XMLA" check box to ensure that the XMLA packages for request and response will be displayed
• Click the Predict button
• Scroll down:
  - the DMX section (if selected) contains the query that was automatically generated
  - the Results section contains the table with the query results
  - the XMLA Request/Response boxes contain the XMLA of the request and response, respectively

Hint: if the Results table is empty, look for an error in the XMLA Response box and please report us the error at mailto:xmlaError@sqlserverdatamining.com

To launch the sample, click here

This sample shows how to use SQL Server Data Mining as a suggestion engine behind an online video store.  The sample uses an Association Rules model built using survey data from Microsoft employees.  The prediction query can be viewed by browsing a special tag near the results in your shopping cart and at the bottom of each item page.

Click here to run the sample

This sample contains a variety of models built upon a dataset that was surveyed from Microsoft employees concerning their movie watching behavior.  Included are several interactive Decision Trees and Naive Bayes models.  The sample uses thin client viewers for data mining models - the DHTML Web Viewers which are available for download as a sample you can run and install yourself.

To start the sample click here

This sample shows how the clustering algorithm can be used to perform automatic data validation through the use of the PredictCaseLikelihood() function.  To exercise the sample, enter values into the form and click the submit button.  If the combination of values has a reasonable likelihood, the form will accept the values.  If not, additional elements of the prediction query indicate the value likely to be unacceptable.  Checking the “Show Details” box on the form will show the query that was sent in addition to the probability ratios used to determine the outlying values.

The source code for this application is available here