Revealing Household Characteristics using Connected Home Products

Abstract

Recent technological advances have helped make our homes more intelligent and responsive to our needs. In this context, analyzing data from smart home products to gain insights is becoming increasingly important. This paper proposes an analytical framework to reveal the characteristics of households using event logs from smart door lock systems. The analytical framework uses constraint satisfaction problems to enable streaming event log analysis to solve the problems of overlapping classes and a lack of information concerning the truth class. The proposed method was applied to two datasets: one consisting of door-lock log data from 40 households and the other of time-use survey data from more than 10,000 households in South Korea. The results verify its effectiveness in terms of estimating the number of occupants in a household. The performance of the proposed method was compared with that of a naive clustering approach in terms of mean squared error.