Between Dinner and Children's Bedtime: Predicting and Justifying Routines in the Home
Nagel, Kristine Susanne
Hudson, James M.
Abowd, Gregory D.
MetadataShow full item record
Much previous research in availability, whether in the office or in the home, has developed linear regression models to help predict appropriate times for interruption. Although these models work well, they tend to be accurate only about 75% of the time. In this paper, we reconceptualize this problem as one of determining routines, rather than availability. We show that the same sensor measures, which predict availability accurately 75% of the time, can predict individual routines accurately 90% - 97% of the time. We argue that better identification of routines can help us to better identify individual availability, as we can develop more tailored models of individual availability in given household routines. In this paper, we also present findings from a day reconstruction method (DayRM) study, which provides more detailed descriptions of three routines in the home: mealtime, bedtime, and leisure.