Characterizing activity patterns using co-clustering and user-activity network

Traditionally human mobility patterns and space activities are studied using recall-based travel diaries. Following the ubiquity of location-based technologies, transportation researchers are revisiting the methods of classifying travel activity patterns using geo-location data. The current study contributes to this research line by leveraging granular and detailed activity information and building individual lifestyle patterns based on top of that. We use 300 days of 402 Metropia navigation app users' origin-destination information to construct an activity-user network. Using the co-clustering method, we discover 16 distinguished clusters or lifestyles in the dataset. The results of this study indicate: (1) Clustering individuals contingent on their similar and dissimilar activities enables us to detect their lifestyle, (2) aggregating the activity space of individuals may misrepresent their lifestyle, and consequently mislead the policies, (3) clustering individuals contingent on their similar and dissimilar activities has the potential to extract the demographic characteristics of individuals, and (4) understanding the human mobility pattern of individuals allows us to create social relationships, and thereby give them an opportunity to share their mobility.