Prediction of state of user´s behavior using Hidden Markov Model in ubiquitous home network

In this paper, we used Hidden Markov prediction tools to predict the state of the behavior of users in a ubiquitous home network. The state of the user´s behavior presents a change of interest in the action of the user. This paper proposes a weight (WEIGHT) for the level of interest in the behavior and the strength of the relation between the behavior and interest, which is the formulation of the user´s interest in the human action. We investigate the feasibility of predicting the next state using the sequence of previously observed states and the action type, and analyze the efficiency of the Hidden Markov Model (HMM). The prediction accuracy of the method is determined. It is found that, on average, the choice of training data leads to a prediction accuracy of 84.61%, while in some cases the accuracy is as high as 91.23%.