CATEGORY: Data Mining & Visualization

REVIEWER: Randall S. Collica, senior business analyst, North America CRM database and business analysis department of Compaq Computer Corporation.

BACKGROUND: Founded in 1982, Compaq Computer Corporation is a leading global provider of enterprise technology and solutions. Compaq designs, develops, manufactures and markets hardware, software, solutions and services including industry- leading enterprise storage and computing solutions, fault-tolerant business- critical solutions, communication products, and desktop and portable personal computers that are sold in more than 200 countries.

PLATFORMS: Compaq Tru64 UNIX and Microsoft Windows 2000. Hardware: Compaq SP700 series workstations running 2 CPU Pentium III processors with 750M of dynamic memory, Compaq ES40 Alpha Server 2 CPU Alpha EV6 running 667MHz with a 350GB RAID array.

PROBLEM SOLVED: The two largest areas where SAS Enterprise Miner has helped us solve particular issues are in customer segmentation and predictive modeling. For customer segmentation, we have typically used the disjoint clustering capability and/or self-organizing maps in order to segment customers or prospects into groups of similar demographics for the purpose of CRM-targeted communications. For predictive modeling, we have used Enterprise Miner to predict customers who are similar to our best set of customers within a market segment. We use these predictions for targeted campaigns. Another application of predictive modeling is for corporate strategic business intelligence. A model was developed that estimates the IT spending of customers or prospects, and the scored results in our customer data warehouse or in the D&B database allowed us to aggregate the estimates in any custom fashion for regional or vertical market analysis, estimates of customer market share and similar strategic business intelligence applications. Depending on the project, the ROI ranged anywhere from two to 50 times.

PRODUCT FUNCTIONALITY: Enterprise Miner functions well as an advanced data mining and business intelligence solution. Another feature we will use in the near future is its text mining for analyzing telesales notes, e-mail text, and customer and prospect Web documents for additional business intelligence applications. Other areas for exploration are trend and sequence analysis in large invoice transaction data and hierarchical clustering for text mining and semi-automated customer profile modeling.

STRENGTHS: Some of the main strengths of SAS Enterprise Miner are the general ease of use in designing and building data mining process flow diagrams with its drag-and-drop node capability, ease in setting up validation and test sets, and the results of those cross validations. Enterprise Miner has a very rich set of algorithms for data mining such as ensemble models and memory-based reasoning. Many other vendors don't have these algorithms, and we have found them very useful. Also, when building clustering or neural network models, the user does not have to recode categorical or nominal variables; it is done automatically. Data manipulation and/or data preparation is easily accomplished within the Enterprise Miner process flow environment.

WEAKNESSES: The naming convention of output data sets is not easy to follow and could be better designed. The node that allows custom SAS code to be written frequently causes an error, which sometimes closes the application and must be restarted.

SELECTION CRITERIA: We had two major reasons for selecting this product. The first was the accuracy and rich set of analytical algorithms along with the ability to easily set up test and validation sets for model training and accuracy. Also, the ability to easily use the rich set of ETL tools already available within SAS to perform data manipulation and data preparation are extremely valuable as these are a good portion of any data mining project. For example, these tools are frequently used to combine, manipulate and enhance data with additional demographics in a customer warehouse.

DELIVERABLES: Enterprise Miner delivers several reports such as interactive cluster profiles, lift or gains charts, tree diagrams and misclassification tables. The entire process flow is captured into an HTML report that includes all settings for each node. This report is great for project documentation and distribution of the results.

VENDOR SUPPORT: Vendor support through their hotline or via the Web is superb compared to many other software vendors. SAS works to solve the problem and monitors progress with a tracking number system. Response is typically within four hours.

DOCUMENTATION: The documentation is generally easy to use and understand, especially the Getting Started section.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access