Bayesian Networks and Small Data

About the Presenter

Benoit Hubert Scientific Director, GfK, France

Abstract

Using Bayesian Networks for analyzing complex data structures is becoming more and more common in the market research industry. Compared to other statistical methods, Bayesian Networks offer a more efficient framework that can quickly reveal the structure, relationships, and dependencies within any domain, even with highly variable data quality and incorporating a wide range of information assets ranging from expert knowledge to massive data manipulation. Even as we are entering the age of big data, the ability of companies to deal with small data in smarter ways will also be key to achieving business success. In this presentation, we will focus on the problem of small samples size adjustment, made possible by Bayesian Networks, from a practical point of view. Using real and simulated data, we will assess the usefulness and effectiveness of Bayesian Networks for small data applications.