This Month's Featured Course
Data Mining Expert Michael Berry Offers New Business Knowledge Course
Michael J.A. Berry has twenty years experience in the field of data mining. With his colleague Gordon S. Linoff, he is the author of the best-selling book Data Mining Techniques for Marketing, Sales, and Customer Support (John Wiley & Sons, 2004), and founder of Data Miners, Inc.Now Berry has teamed up with Linoff again to offer the Business Knowledge Series course, Introduction to Survival Analysis Using Empirical Hazards. The course introduces survival analysis in the context of business data mining and focuses on understanding customer behaviors that have a time-to-event component. Berry recently took time out from his busy consulting schedule to answer a few questions about the course.
Q. What job roles would benefit most from attending this course and why?
A. Anyone whose job involves analyzing large quantities of customer data will benefit from this class. Survival analysis is one more tool that these analysts can add to their tool box. It provides a new way of looking at questions of customer value, customer loyalty, customer retention and repeat purchase behavior.Q. What is unique about your approach to this course topic?
A. One thing that distinguishes our approach to the topic of survival analysis from all others we have seen is our exclusive focus on business topics and business data. In fact, a better name for the course might be Analyzing Time-to-Event Problems in Business or Applications of Survival Analysis to Real Business Problems. It can be very hard for students to translate what they learn in a class based on clinical trials data to the very different world of customer relationship management. The sheer size of a typical corporate customer list with millions of customers is so different from the few dozen or few hundred patients in a clinical trial that the quantitative difference becomes a qualitative difference. The techniques that make sense for small data do not scale well to big data. Conversely, the empirical hazards method we teach for big data would not work on small data sets. The data used in our class tracks mobile phone subscribers and repeat visitors to a Web site.But the thing students like the best about our classes is that we bring many years of hands-on practical experience into our classroom lectures. We are practicing consultants who get our hands dirty mining data every day. This makes it easy for students to connect theory and practice because we have lots of stories and examples of how we have applied the techniques we teach about.
Q. What specific tools will students gain from attending this course?
A. There are four key skills students can take with them:- How to make quantitative comparisons between different customer segments.
- How to calculate customer value in a way that takes expected remaining tenure into account.
- How to forecast future customer population counts in subscription-based businesses such as long-distance telephone, mobile telephone, newspaper, ISP, magazine, Cable TV, etc.
- How to model repeated events such as purchases, Web site visits, pre-paid phone card recharging, etc.
Q. What specific business pains are addressed by this course?
A. Subscription-based businesses are always being surprised by sudden spikes in customer churn or cancellations. They have to manage these spikes reactively. They shouldn't have to. Using survival analysis, these spikes can be forecast and managed proactively. These same businesses have a hard time predicting what the effect of a change in business rules such as a change in the minimum credit limit or a change in marketing channels will have on customer value and customer retention. Survival analysis provides the tools for "what if" analysis. Non-subscription businesses also have many time-to-event problems such as predicting time to next purchase or time to next Web site visit.Q. What tangible results do you hope your students will see?
A. Better understanding of customers. Greater ability to take the time dimension into account when modeling business processes. Nearly all business problems have a time dimension, but it often gets ignored because it is difficult to deal with using standard tools such as regression. Survival analysis lets you stop ignoring time.View complete course outline for details.
View schedule and register.