Using the SQL Server Data Mining Algorithms: Tips and Resources

Article
1/17/2024

Work in progress

****
This article is a list of frequently asked questions about each of the data mining algorithms provided by SQL Server Data Mining.
Resources are also provided, where available, to help you understand how these algorithms compare or align with other open source or commercially available methods, including Excel features and third-party plug-ins for Excel.

What algorithms are included with SQL Server?

The algorithms (and other features) that are available depend on which version you have. For complete details, see this page:

SQL Server 2000
- Clustering
- Decision trees

SQL Server 2005
A huge number of new features were provided including these new data mining algorithms:
   - Decision trees with regression
   - Association rules
   - Sequence clustering
   - Time series (ARTXP)
   - Neural network
   - Naive Bayes

An easy Excel client for data mining was also added

SQL Server 2008 and SQL Server 2008 R2
   - New ARIMA model for time series
   - No other algorithms, but cross-validcation was added, as well as the ability to split data into a traingi and testing set.
   - Added the ability to filter models while using the same minign structure. This is

The Excel client was updated to support all new features.

SQL Server 2012
No change in data mining features, but the client was updated ot support 64-bit Excel.