Event Prediction in an IoT Environment Using Naïve Bayesian Models

Abstract

In many Internet of Things (IoT) scenarios, there is a need to predict events generated by objects. However, because of the dynamicity of IoT environments, it is difficult to predict with certainty if/when such events will occur. Probabilistic reasoning allows us to infer dependent probabilities of events, from other events that are either easier to detect or to predict. In this paper we propose an architecture that employs a Bayesian event prediction model that uses historical event data generated by the IoT cloud to calculate the probability of future events. We demonstrate the architecture by implementing a prototype system to predict outbound flight delay events, based on inbound flight delays, based on historical data collected from aviation statistics databases.