Exploring time diaries using semi-automated activity pattern extraction

Identifying patterns of activities in time diaries in order to understand the variety of daily life in terms of combinations of activities performed by individuals in different groups is of interest in time use research. So far, activity patterns have mostly been identified by visually inspecting representations of activity data or by using sequence comparison methods, such as sequence alignment, in order to cluster similar data and then extract representative patterns from these clusters. Both these methods are sensitive to data size, pure visual methods become too cluttered and sequence comparison methods become too time consuming. Furthermore, the patterns identified by both methods represent mostly general trends of activity in a population, while detail and unexpected features hidden in the data are often never revealed. We have implemented an algorithm that searches the time diaries and automatically extracts all activity patterns meeting user-defined criteria of what constitutes a valid pattern of interest for the user’s research question. Amongst the many criteria which can be applied are a time window containing the pattern, minimum and maximum occurrences of the pattern, and number of people that perform it. The extracted activity patterns can then be interactively filtered, visualized and analyzed to reveal interesting insights. Exploration of the results of each pattern search may result in new hypotheses which can be subsequently explored by altering the search criteria. To demonstrate the value of the presented approach we consider and discuss sequential activity patterns at a population level, from a single day perspective.