Joint-Activities Generation among Household Members using a Latent Class Model

Abstract

Individuals’ decision to include an activity in their daily agendas is based on their needs. They take decisions of activity participation, considering the activities and schedules of other household members. This results in activities that involve the joint participation of household members. Limited modelling research is available on the notion of the generation process of joint activities of household members. Besides, the time span for most of the operational models is a single day. The current paper’s focus is on modelling joint-activity generation where the time span is based on multiple days (one week). We used a latent class model (LCM) and estimated it using the data from the National Travel Survey (NTS (UK)) for the years 2013-14. On a given day, it is assumed that the individual includes an activity (that may be joint activity) in his agenda based on his need. The model predicts the probability of alternatives for an individual. Each alternative represents a series of binary outcomes (i.e. 1 or 0) in relation to the generation of a specific type of joint activity for every single day of the week. Four latent classes are identified based on model-fit parameters: frequent joint-activity performer, weekdays joint-activity performer, non-frequent joint-activity performer, and weekend joint-activity performer. The results show that both household and individual attributes impact the individual’s weekly joint activities generation. The individual’s economic status (i.e. full-time worker, part-time worker, etc.), available time window after mandatory activities, number of household members and life cycle stage (structure of the household, i.e. single-parent family etc.) affect which latent class the individual belongs to. Future research is focused on estimating the number of tours, and travel parties involve in joint activity models. It also focuses on integrating the model in some functional activity-based model.