Data Mining Services

Data Mining Services is perhaps MicroStrategy’s best kept secret.  It’s a powerful set of tools and functions built right into the product that offers advanced statistic report capability including predictive analytics, market basket analysis, networking and trending.  This functionality is free and is included out of the box.  There’s no additional software needed and no special permissions to use it.  Since it’s not sold separately, it’s rarely advertised, though it truly is one of the most powerful components in MicroStrategy.  You don’t need an advanced mathematics degree to use it (though of course that helps), so give it a try and see what you can build.  Everyone loves looking at predictive reports, so if nothing else you could use some samples to drum up excitement about your project.
Full disclosure: I’ve produced a few predictive reports and attended MicroStrategy’s Education Class on the subject, but I’m definitely no expert.  Instead of trying to re-document it here, I’m going to defer to the fantastic documentation that MicroStrategy already provides.  Hopefully this will serve as a good way to unearth some of these hidden features and will get you on your way.




Predictive Metrics
The most popular feature and the very definition of Data Mining Services are Predictive Metrics.  These are special types of metrics that will carry the trends of your existing warehouse data into the future based on various statistical models.

MicroStrategy TN20000

To create your own, you first create a Training Metric which is where you will define your model.  Once created, you run it in a report related to the model you defined which will spit out your shiny new Predictive Metric that you can now use on normal reports.  That’s the gist, for more details, check out this very thorough tech note that will show you how to build the prediction pictured above.

Market Basket Analysis
This style of report is almost required for anyone in the retail space.  It finds relationships between items and scores them on how strong they are.  The most classic example is “Beer and Diapers“, which basically states that those two products have a high affinity for each other because young dads who go out  on a diaper run are likely to pick up a six-pack apparently.  With that insight, a store could put those items close to each other and increase impulse sales.

For more information on setting up Association Models, the statistical model behind the Market Basket Analysis, check out these tech notes.

Integration with Other Tools
MicroStrategy can generate most of the common types of statistical models, but can actually score a great deal more.  If you have a more advanced tool for generating statistical models, you can import it into MicroStrategy for scoring using PMML.  MicroStrategy doesn’t position themselves as a competitor to other tools but rather a complement.  The model generation they offer is so that you can still make use of the tool and get started without owning more complex systems, though if you do they make every effort to help integrate those into your BI solution.

Function Plugin Wizard
This also falls under the Data Mining Services I think, because I can’t really think of another reason you’d ever use this.  Basically, instead of using an ApplySimple or ApplyAgg to use a database function that MicroStrategy doesn’t natively support or one you’ve built yourself, you can wrap it in a native function and use it just like any other function in the system.  You can do this using the Wizard which is a plugin with Microsoft Visual C++ without writing any code.  Of course, you can get really fancy and add Analytical Engine handling for your custom function, but that will require some coding.

If you’ve ever accidentally selected this option during an install, you probably got an error about missing components.  Installing it is actually pretty tricky, and there’s a rather lengthy tech note on setting it up.

Further Reading
For comprehensive documentation on everything Data Mining Services has to offer, check out Chapter 14 of the Advanced Reporting Guide.

You may also like...