Context recognition in a smart home: living experiment

Smart environments have emerged as a thriving field of research. Understanding user's context is one of the core areas of smart home research. In this study we present a smart home living experiment in which a test subject lived two weeks in the EleHome lab. During this time sensor data was gathered along with annotation data. We used naïve Bayes classifier in a supervised learning setting to recognize a few of the basic contexts and to validate the gathered data. The first results were good for simpler contexts, but further analysis is needed for more complex activities. All in all the study gives us a good baseline for further development of context recognition methods.