Human behavior inside a home strongly affects indoor air quality, based on a study in sensors-rich “smart homes” environments. Researchers at Washington State University measured air pollutants (PM, ozone, CO2, and 13 organic compounds) inside two home fitted with a variety of sensors to detect human presence and activities, such as motion, ambient light, window and door openings, and temperature. The relationship between activity patterns and indoor air pollutants was analyzed through machine learning techniques (e.g. decision tree learning). Among the many results of the analysis, measures of temperature are strongly related (not necessarily in a causal way) to PM2.5 indoors; methanol with laundry, food activities, and human presence in a room. These first results may be used to continue developing predictive relationships for exposure assessment and for ventilation recommendations.