Personalized multi-activity scheduling of secondary activities

Abstract

Understanding how the individuals from a certain city or region plan their activities is very useful for many applications, such as travel demand modelling, land use planning and redevelopment, or market research and analysis. Accurate multi-activity scheduling is a non-trivial problem for its number of dimensions, and the amount of information involved in human decisions. The concept of fixed and flexible activities allows to study this problem at two levels. It can be assumed that mandatory activities like rest, work or study, are already prearranged, and non-mandatory activities like shopping, eating or running errands are planned on-the-fly. This paper proposes an algorithm that emulates personalized multi-activity scheduling of flexible activities within well defined time windows. The algorithm uses a mental map composed by an Activity agenda and a Set of known places obtained from real data. It calculates the maximum utility activity-trip chain according to temporal and spatial conditions of the decision. It solves the number of the activities to be performed, the sequence, the locations, the start times and the activity durations. Transportation modes of the trips between consecutive activities are also predicted. This method (i) doesn’t prioritize or fix any scheduling dimension, (ii) uses common available data (i.e. travel surveys and land use datasets) as its input, (iii) generates personalized solutions, and (iv) can be configured to schedule flexible activities of a full population in tractable times. In order to show the efficiency of this approach flexible activity patterns from Singapore were extracted from a travel survey carried out in 2012, and the method was tested with a fraction of the survey observations which were not used for the estimation of the models. Five types of flexible activities were included. Similar shapes of general travel time and activity duration distributions between real and predicted activity-trip validate the potential of the proposed algorithm. Finally, the method was used for within day re-planning of a synthetic population using MATSim, a well known agent-based transport simulation platform. Results show that the algorithm is computationally feasible for large-scale scenarios.