Leveraging Analytics for Big Advantage in a Big Data World

With increased demand for mass customization and personalization, the emergence of Web 2.0, and one-to-one marketing, and the need for better risk management and timely fraud detection, the pressure is on for organizations to improve their ability to extract, understand, and exploit analytical patterns of customer behavior and strategic intelligence.  In a new SAS e-learning course, Big Data Quarterly columnist Bart Baesens, Ph.D., who is also a professor at KU Leuven (Belgium), and a lecturer at the University of Southampton (United Kingdom), will present interactive lessons covering how to successfully adopt state-of-the art analytical and data science techniques for advanced customer intelligence applications.

Baesens recently explained why it is imperative for organization to stay on top of the latest analytical trends and approaches. “It is not our aim to provide exhaustive coverage of all analytical techniques previously developed but rather to discuss the ones that really provide added value in a business setting,” Baesens noted. 

BDQ: What are the newest trends in analytics which you cover in the course?

Bart Baesens:  Based upon some of our recent research findings, we extensively zoom into social network learning.  We discuss how to build a social network, including how to define the nodes and edges, and how to use it subsequently for building analytical models in fraud detection, churn prediction, and response modeling.  We also explain how complex, yet sophisticated, analytical models based upon neural networks and Support Vector Machines can be used in a white-box, explanatory way using rule extraction techniques and two-stage modeling.  The topic of survival analysis is also covered extensively.  We currently see a lot of interest in this in the industry in terms of how to predict the timing of events of interest such as churn, default, and fraud. The course concludes by reviewing some key concepts and recent research insights to monitor, back-test, and benchmark analytical models.  Throughout the entire course, we inject insights from our recent research and consulting experience.  

BDQ: What are the key advantages of a self-paced e-learning course?

Bart Baesens:  We spent about 4 months designing the course.  The content has been carefully reviewed and the videos edited.  Each video is subtitled, and can be paused and replayed as many times as desired.  In addition, the participant gets access to all course material for 1 year, so that the entire course can be taken at the participant’s own pace and customized sequence.  If you are building a random forest analytical model, for example, you can quickly refresh the technique’s workings by selecting the appropriate video. 

BDQ: How does it compare to a traditional classroom course?

Bart Baesens: A traditional classroom course has to be scheduled on a particular date at a particular location. The e-learning course corresponds to a 3-day classroom course.  We have taught it as a classroom course worldwide more than 100 times.  However, due to budget constraints and busy work agendas, it’s getting more and more difficult for business people to travel and commit to a 3-day classroom course.  The e-learning course fills this need, since travel is no longer needed and the participant can take the course wherever and whenever they want. They only need a laptop, iPad, or iPhone with an internet connection. 

BDQ: Who is this course targeted at?

Bart Baesens: The  course is targeted at the business professionals such as data scientists, data miners, quantitative analysts, analytic model builders, analytic model auditors/validators, and consultants  who are working in industries such as finance, retail, insurance, pharmaceutical, manufacturing, and government. The focus of the course is not on the mathematics or theory, but on the practical application for risk management, customer segmentation, fraud, and churn. It is not our aim to provide exhaustive coverage of all analytical techniques previously developed but rather to discuss the ones that really provide added value in a business setting. 

BDQ: How is interaction built in?

Bart Baesens:  We incorporate interaction in two ways.  Between the videos, we include short review questions to test if the participant understands the concepts discussed.  Every chapter also has a review quiz which typically consists of about 10 multiple choice questions.  Each question has extensive feedback about why certain options are wrong or correct.  A certificate is provided upon satisfactory completion of all end-of chapter quizzes.

BDQ: What are some industry examples that you have included as reference points?

Bart Baesens: Given the relevance of big data and analytics in today’s business environment, we try to include as many industry examples as possible.  More specifically, we discuss short case studies in credit risk management, churn prediction, response modeling, fraud detection, market basket analysis and customer lifetime value modeling to illustrate how analytics can help create added business value and identify new strategic opportunities. 

In addition to Baesens, the course is presented by Christophe Mues, Ph.D., professor at the School of Management of the University of Southampton (UK); or Wouter Verbeke, Ph.D., assistant professor, Business Informatics, University of Brussels (Belgium); or Thomas Verbraken, financial industry research associate, KU Leuven (Belgium).

For more details on the course, titled, Advanced Analytics in a Big Data World, go here.

Image courtesy of Shutterstock. 

Related Articles

Translated to an analytical setting, Ockham's principle, also known as Ockham's razor, basically states that analytical models should be as simple as possible, free of any unnecessary complexities and/or assumptions.

Posted April 08, 2015

These days, everybody talks about big data and analytics but does anybody really know how to tackle it? Everybody thinks the others are doing it and hence also claim to do it as well, but it's likely that only the fortunate have had (positive) experience(s) with it so far.

Posted June 26, 2015

A typical organization loses 5% of its revenues to fraud each year, and in the U.S. alone the total cost of (non-health) insurance fraud is estimated at more than $40 billion per year, according to Bart Baesens, one of the authors of a new book on data science techniques for fraud detection. While opportunities for fraud are continuously evolving, fraud detection always has the biggest impact if it is done quickly. Fortunately, says Baesens, big data analytics can help to make early fraud detection possible.

Posted August 05, 2015


Subscribe to Big Data Quarterly E-Edition