A Model for Predicting IoT User Behavior Based on Bayesian Learning and Neural Networks

Abstract

To facilitate the allocation of energy and resources in the Internet of Things system, this paper presents a model for predicting user behavior in Internet of Things environments. The model is based on Bayesian learning and neural networks and is designed to provide insights into the future behavior of users, allowing for the allocation of resources in advance. In this paper, the data are preprocessed by data merging and format processing, and then the association rules are mined by association rules analysis. Finally, the data are utilized to train the behavioral prediction model of the short‐duration memory network via Bayesian optimization. The experimental results showed that the average running time of the research model was 1.682 s, the average accuracy was 96.77%, the average root‐mean‐square error was 0.382, and the average absolute error was 0.315. The designed behavior prediction model is capable of effectively predicting the user behavior of the Internet of Things, thereby enabling the reasonable allocation of energy and resources in the Internet of Things system.