Micro Reality Mining of a Cell Phone Usage Behavior: A General Bayesian Network Approach

Abstract

Successful products and services result from a keen awareness of the micro motives and factors underlying consumer behavior in the real world. Similarly, the success of ubiquitous decision support systems is heavily dependent on the capability of micro-reality mining from consumer behavior data. Micro-reality mining is defined by the system's ability to extract a set of trivial but meaningful rules of action from consumer behavior data. This study is based on the hard fact that such trivialness leads to the macro behavior of consumers. As an example of successful micro-reality mining, this paper proposes a new method based on General Bayesian Network (GBN). Using MIT students' real life data, we applied GBN and obtained a set of causal relationships among a set of relevant variables. The what-if and goal-seeking simulations with the causal relationships supported by GBN allowed us to explore the usefulness of GBN-driven micro-reality data mining.