Novel model for inhabitants prediction in smart houses

Purpose ‐ The purpose of this paper is to present a novel model for inhabitant prediction in smart houses based on daily life activities. It uses data provided by non intrusive sensors and devices to predict the house occupant. The authors' model, named Behavior Classification Model (BCM), applies Support Vector Machines (SVM) classifier to learn the users' habits when they perform activities, and then predicts the user. BCM was tested using real data and compared with a frequency based approach. In this paper the authors present also their approach to improve the accuracy of BCM using SVM feature selection algorithm. Design/methodology/approach ‐ The model, named Behavior Classification Model (BCM), applies Support Vector Machines (SVM) classifier to learn the users' habits when they perform activities, and then predicts the user. Findings ‐ BCM was tested using real data and compared with a frequency based approach. In this paper the authors' also present their approach to improve the accuracy of BCM using SVM feature selection algorithm. Originality/value ‐ The paper is based on blind user recognition in smart homes.