Routine activity location choice

The destination choice model in POLARIS is a multi-nomial logit model with zones as alternatives and accounts for competition and agglomeration effects. For runtime purposes, it employs stratified sampling based on travel time and employment, with 4 strata and 200 alternatives each. The choice of individual location in a zone is done randomly. The location of a work or school activity is usually a long-term choice and, if a person performs this routine activity on the modelled day, dominates the daily activity planning. POLARIS models the location of regular work or school activities before considering other activitites. Since a multi-nomial logit model generates choice probabilities per zone there is no guarantee that the outcome of the model for all agents agrees with the exact distribution of attractions. POLARIS therefore keeps track of the number of simulated employment within a zone and makes zones unavailable once all employment is allocated.

For school locations, children under the age of 14 always choose the home location zone for school if there is a school available. If not, the model looks for all schools within a certain driving distance (10 minutes for children under 14, 30 minutes for children aged 14 to 18 and 60 minutes for children over 18) and then randomly picks on based on the (relative) number of schools in a zone.