Optimization of Real-Time Experience Measurement with Bayesian Networks
Presented at the 4th Annual BayesiaLab Conference in Nashville on September 29, 2016.
Benoit Hubert
Global Scientific Director, GfK, France
Abstract
Bayesian Networks are beautiful tools for structure learning and knowledge discovery. Once a model has been adjusted, they are able to deploy inferences across all variables when some of them change (a posteriori update - instantiation).
In the Market Research Industry and in data enrichment in general, questionnaire time reduction is a key issue. For GfK clients, improvement of response rates, less fatigue, cost reductions are well-known benefits when significantly reduced questionnaire length is achieved.
We will show in this presentation how we have leveraged Bayesian Networks in a real client project in combination with optimization methods. And finally, a special focus will be presented on the management of real-time constraints and the delivery of reliable KPIs.