Analysis of buildings occupancy has become a relevant research field following energy management and smart grid trends. Occupancy can be associated with energy needs, and therefore a good knowledge of occupancy behavior can substantially help to manage better energy sources and to optimize overall consumption. In this work, we analyze the use of commonly employed occupancy measurements under the residential scenario. We present the design methodology, the implementation, the experimental tests and occupancy analysis of a multi-sensor measurement prototype. We propose a signal processing methodology based on Haar wavelets and multivariate analysis using the Normalized Cross-Correlation and Principal Component Analysis to improve the linear regression model of occupancy.