Software

Information systems and the enormous amounts of data they produce have become a central part of modern corporate life. However, the abundance of data is starting to blur the picture and in many cases is bogging down our capability to identify relevant information. Many statisticians and analytic professionals spend a big part of their time performing the tedious tasks of data preparation and variable selection. At KDD 2004 in Seattle KXEN is launching a new productivity tool called the KXEN Companion. As the name suggests, this product is designed to accompany the early phase of model creation, allowing the user to access a large data set with thousands of variables, encode the data and automatically determine the variable contribution. You can assess the resulting lift, select the desired variables and take them directly into your SASTM environment to train a model.

How does KXEN help?

KXEN accelerates the modeling process by providing quick answers to key issues:

• Data Encoding: KXEN Companion delivers two datasets containing the binning and the banding for continuous, nominal and ordinal variables. It allows you to quickly encode and assess the contribution even of difficult variables such as zip codes, first names, telephone numbers etc..
• Correlated variables: KXEN helps select the most important variable, and may suggest to keep or to discard several.
• Missing Data: KXEN temporarily creates internal value replacements to assess variables that contain missing values.
• Speed and Reliability: KXEN is very fast and is used by industry leaders worldwide. KXEN companion provides powerful metrics for quality and robustness of each variable as well as the entire selection.
• Fast Fail: With KXEN Companion, you will know right away, if a dataset is worth spending your time on.

As an analytic professional KXEN Companion will help you to increase your productivity, allowing you to build more models faster. You can be confident that you have exhausted the full potential of your available data, leveraging your firm�s data warehouse investment. This in turn will allow you to concentrate on more important issues such as model tuning and improving the overall results. KXEN Companion is also a great vehicle for communication with the business side. You will be able to use all the data you have collected and can quickly point out variables that are important to gather or to collect.

Dr. Richard De Veaux, Professor for Mathematics at Williams College, MA will hold a series of web seminars to share his experience with KXEN Companion. To register for one of these sessions, please follow the link below.

http://www.kxen.com/workshops/kxen_companion.php