STATISTICA Blogs

Bob Nisbet is a Data Mining consultant and a Data Mining Instructor. He uses STATISTICA as a tool when building models for companies in Telecommunications, Insurance, and Banking. He is also a senior author of the book, Handbook of Statistical Analysis and Data Mining Applications.

• Tell us about your education and career.
My background is in the Biological sciences, with a BS in Forest Management (including a lot of statistical analysis), MS in Ecology (with a lot of ecological statistical analysis), and Ph.D. in Ecosystem analysis (that led to a lot of simulation and modeling). I taught Ecology and Botany at Malone College in Ohio, and I taught Environment Science and Resource Management at UC Santa Barbara. I have worked professionally in environmental impact analysis and modeling, leading the Vegetation and Rare Plant studies for the large environmental impact statement for the U.S. Air Force-proposed MX-Missile program. I was invited by the Russian Academy of Science to participate in a 3-week environmental evaluation of oil and gas extraction lands in NW Siberia in 1993 (just after the fall of Communism). At UCSB, I built simulation models for forest growth under simulated global warming conditions 90 years in the future for Michigan, Minnesota, Alaska, Costa Rica, and NE Siberia. In 1994, I was challenged by another forest modeler to come to AT&T to help him develop the new science of Data Mining. I became fascinated with the practice of data mining in business; I left biological science behind, and never looked back. I led the NCR team to develop the first data mining models for CRM in SAS and Clementine in 1998. I have taught numerous classes on data mining tools such as training classes for STATISTICA Data Miner. I published two data mining tool reviews for DM-Review. I have worked in many industries including Telecommunications, Insurance, Banking, and membership clubs (AAA). Much of this experience and background is reflected in my service as the senior author of the book Handbook of Statistical Analysis & Data Mining Applications (Academic Press, 2009). Currently, I and my author team are under contract to write a companion book on text mining.

• How do you use STATISTICA?
In the past, I used STATISTICA Base for statistical analysis in biological sciences. Since about 2004, I have used STATISTICA Data Miner for much of my modeling activities and as an entry point to the base statistical analysis routines. Sometimes, I use the powerful spreadsheet capabilities of the tool for data analysis and preparation purposes apart from data mining. The graphics capabilities are what drew me to STATISTICA in 1989; I have been using them ever since to display charts and graphs in my publications.

• What are some interesting consulting projects that you’ve used STATISTICA for?
I have built Underwriting loss risk models with STATISTICA; I have used the time-series algorithms for analyzing whaling ship sightings of sea ice in relation to latitude. I built a customer acquisition model for commercial customers of a medium-sized bank.

• What types of customers do you usually consult with?
My specialties are in the CRM modeling of customers in Telecommunications, Insurance, and Banking.

• What advantages do you see in STATISTICA compared to other general tools?
The two most important capabilities of STATISTICA that I rely upon are:
1. The 100% integration between all the tools. This means that the powerful statistical analysis and graphical tools are always available for reporting results of models using any modeling algorithm.

2. The exquisite automatic modeling capabilities of several important algorithms, like STATISTICA Automatic Neural Networks (SANN), and more importantly, the automatic implementation of ensemble modeling and results presentation in the Data Miner Recipes interface. These automated features provide an important part of a data mining “dashboard,” which permits novice users to operate successfully the complex technology “under the hood” with a minimum of training.

• What are STATISTICA’s strengths?

1. An extremely rich selection of modeling algorithms, graphical display tools, and reporting methods;
2. 100% integration between all modeling and display features of the tool
3. The high degree of automated action available in the tool
4. Relatively low price, compared to other mainstream data mining tools
5. Excellent customer support
6. Excellent online help facility
7. Excellent help facility in the tool, combined with the excellent presentation of help topics in the included electronic encyclopedia of data mining.

• Would you recommend STATISTICA to a colleague or friend? Why?
I have recommended STATISTICA for the tool of choice to a large insurance company. All members of the Analytics team agreed with my choice. I continue to recommend the tool to companies that are not entrenched with SAS. In spite of these obvious advantages over SAS, companies that have used various SAS tools for a long time appear to identify their analytical operations with the tool, and it is extremely hard to displace it for business reasons rather than technical reasons.