Data Mining Options for the UNT Community

By Patrick McLeod, Research and Statistical Support Services Consultant

Even before our economy took the downward turn that we’re currently wading through, corporations, institutions and organizations of all kinds were investing non-trivial amounts of resources in improving efficiencies. Whether it’s called “reducing churn” or “realizing efficiencies” or some other terminology, the same group of techniques lies at the heart of these ideas: Data mining.

There are several data mining options available to the UNT research community that, depending on your classification (full-time faculty, academic, etc.), might be a good fit for your data mining needs. The two primary data mining tools that I recommend to folks asking for advice here in the office are the SAS suite and Weka.

The SAS suite offers a set of tools that are widely used in the business world for a variety of statistical and analytical needs. The is one particular tool in our SAS suite that pertains to data mining that has the lion’s share of the data mining market in the business world: Enterprise Miner. The current version of Enterprise Miner is 5.3. It is available to those users who qualify for the Teaching and Research license for SAS at UNT (full time faculty and staff engaged in academic teaching and research activities and sponsored graduate students). For more information about the qualifications for a sponsored graduate student, please see the following article from the April 2008 edition of Benchmarks Online. Enterprise Miner is currently not a part of our Administrative license.

Using Enterprise Miner does come with some non-trivial overhead, though. In order to install Enterprise Miner on a computer, the user (or the user’s support staff) will need to install a whole host of SAS services and platforms such as SAS Analytics, SAS Metadata Server, SAS Management Console, etc. All of these services are tied to the user account on your Windows machine and some of them will require administrative level access in order to function properly. I’ve been supporting SAS at UNT for over six years now and I still run into issues with the web of dependencies that make Enterprise Miner tick. Another thing to consider with Enterprise Miner is that the current version of it is not compatible with Windows Vista in any version nor is it compatible with Windows XP Pro. You can find out more about installing Enterprise Miner here: http://support.sas.com/documentation/onlinedoc/miner/install53.pdf .

A powerful open source machine learning software that is offers some of the functionality of Enterprise Miner that I am spending more time with and expanding my training on is Weka. Weka is written in Java and is a project of the University of Waikato in New Zealand. Weka has a nice wiki and since it is written in Java, it can be installed on a wide variety of platforms as long as you have the necessary version of Java.

I am in the early stages of using Weka, but at this stage I would say that it is a very effective alternative to machine learning approaches that some researchers might use in SAS Enterprise Miner and without the overhead. I hope that some or all of you have found this rundown of interest and, as always, if you have any questions about Enterprise Miner or Weka, please email me at Patrick.Mcleod@unt.edu. Until next time, happy trails!