Click the DOWNLOAD link below to download a zip file that includes a C# project with the full source code for the scenario described in the Programmatically generating prediction queries for nested tables" tip, as well as a SQL Server 2008 backup of the source database (Test.bak).

This sample allows you to explore the Time Series enhancements in SQL Server 2008. It provides readily-accessible time series data, an easy-to-use interface to model and play with the algorithm features and an explanation window with a link to the server-side DMX queries so you can understand what's going on under the hood.

Prerequisites

• A connection to SQL Server 2008 Analysis Services (Enterprise). A 180-day evaluation is available for download.
• SQL Server 2008 Client Components installed on your local machine.

Installing the sample

Just unzip the files to a local folder and run DMTimeSeriesSample.exe to launch the sample. You can click the Help button on the toolbar for help with the feature set.

More Information

This article provides more details about the sample and the SQL Server 2008 Time Series features that it highlights.

This zip file contains a Visual Studio solution with the C# source code for building a class assembly for the CreateDataSource stored procedure.

The AccuracyEx.EvaluateModelingData stored procedure produces the data for rendering a chart that shows the training set size affects accuracy.

 Excel 2007 format version of the WBCD data set useful for exercising the Data Mining Addins for Office 2007

 Original dataset is available from :

http://www.ics.uci.edu/~mlearn/MLRepository.html

A few bugs were detected in the Plug-in Algorithms sample. This file contains the updated source files for this sample, with the fixes for all the known issues

This package includes the code and supporting files for the Lack of Fit stored procedure and sample application. See the included readme.txt file and the article "Calculating Lack of Fit for Microsoft Data Mining Models" for complete details.

This file is a template you can use to create datasource objects using SQL Management Studio(SSMS).  This allows a SSMS workflow such as: 

1: Create a database by right-clicking on the Databases folder
2: Create a datasource using this template
3: Create a mining structure using CREATE MINING STRUCTURE
4: Train structure using INSERT INTO and the datasource just created
5: Add models
... 

To install this template so that it shows up in Template Explorer, save it under C:\Program Files\Microsoft SQL Server\90\Tools\Binn\VSShell\Common7\IDE\sqlworkbenchprojectitems\AnalysisServices\XMLA\Management.  You will have to restart SSMS before it will appear in the explorer.  Alternati

This file contains an Excel AddIn to perform forecasting as described in chapter 16 of Data Mining with SQL Server 2005 by Tang and MacLennan

Visual Studio 2005 is required on the machine where the sample will be installed

Installation steps:
- extract the content of the archive file into your IIS root folder. This folder is usually c:\inetpub\wwwroot
- from Control Panel\Administrative Tools, launch the Internet Information Services management console
- expand the Web Sites\Default Web Site node
- right-click on the newly created ArtOfClustering folder and select Properties
- in the Application Settings section of the properties wizard, click on the Create button, effectively creating a web application
- in the web.config file of the application, change the connection string to point to a valid Analysis Services server and a database that has permissions for the credentials used by ASP.Net  
If your web site is currently configured to use a previous version of ASP.Net (such as ASP.Net 1.1) you will need to open the web application inside Visual Studio 2005 . This will adjust the web application settings to use ASP.Net 2.0

The 2-dimensional cluster test application generates ellipses based on parameters you specify - the center, width, height, angle and density and then builds a session mining model to cluster the generated points.  The result is the original data points colored by the clusters found by the data mining algorithm.

The application is great for experimenting with the clustering algorithm and seeing how the cluster parameters impact the end result

The DMPredict sample implements a VBA function to create a query against a mining model using data in a user-selected range.  To use this sample as provided, you first need to restore the supplied AS database backup using SQL Management Studio.

The syntax of the DMPredict function is

DMPredict(string database, string modelname, string DMXFunction, range inputrange)

For example, in cell G2 of the spreadsheet you could put the following code to get the cluster label for the row:

=DMPredict("ExcelDemo","XLClusters","Cluster()",B2:F2)

Or you could put the following in F9 to pr

This sample contains the source code and compiled executable of an application that embeds and displays the data mining viewers of SQL Server 2005.  Using this application you can load two different models and compare them side-by-side in a splitter control.  Details are in the enclosed README.

This sample provides a new data mining viewer accessible from SQL Management Studio and the BI Dev Studio.  The viewer works with models built with any algorithm helps you understand how the model content is organized by using a tree control to navigate the hierarchy and a browser window to display the content of each individual node.  All source code is included and installed in the path you specify.

After installation, access the viewer by browsing a model and selecting "Generic Content Tree Viewer" from the viewer drop-down at the top of the viewer window. (Updated for September CTP and RTM versions of SQL Server 2005)

This sample shows how the clustering algorithm can be used to perform automatic data validation through the use of the PredictCaseLikelihood() function.  It contains all the source code as well as a .dmx file containing the PMML for the clustering model used on this site.

You can launch the sample and see it in action from here.

To install the application, follow the instructions in the included readme.rtf file.

.

Each XML file in the download can be loaded into the DMSample query tool (available as a separate download) and contains the model creation, training and query statements. Note that these samples were originally written for SQL Server 2000 DM and therefore do not include SQL 2005 features.

 DMSamp.exe is a sample VB application, originally written for SQL Server 2000, that can be used as a simple query tool for SQL Server Data Mining. It  can be used with a SQL 2005+ Analysis Server as well but it does show not SQL Server 2005 Data Mining constructs like the mining structure, datasource and DSV. However, you can send any DMX statement and retrieve both flattened and hierarchical results.

This package includes the following files:

1. DMSamp.exe - the application executable
2. DMSamp10.exe - a self-extracting executable containing the VB source code for DMSamp.exe
3. MSHFLXGD.*- files for the Microsoft FlexGrid control used by DMSamp.exe
4. msmdxtb.dll - QueryPane control used by DMSamp.exe
5. readme.txt

AMO (Analysis Management Objects) is the new object model in SQL Server Analysis Services 2005 for creating and managing Data Mining and OLAP objects. AMO is a CLR-based model and is accessible through any of the .NET languages like C# and VB.NET. This short code sample in C# shows you how to programmatically create SQL Server 2005 Data Mining objects using AMO.

 
I will start with a description of how the server handles attribute values and how they are modeled. After that, I will explain how to obtain the attribute values in different contexts. Also, the attached file contains some helper functions. I will explain the usage of the code in this file which, in some form, will be eventually part of the plug-in tutorial. A preliminary version of the helper classes I'm attaching is already part of the tutorial, the DataValues.h file.

Attributes can basically be of 2 kinds: continuous and discrete. Discretized attributes are a discrete view over data that originally was continuous. Discrete attributes have a number of distinct states. The actual values for the states are not of importance for the algorithm, these values do not play any part in modeling. They can be considered State1, State2 etc. A distinct state (which is usually 0 now, but is not guaranteed to have this