Sensitivity analysis of activity scheduling parameters with a parameter optimization framework

Abstract

Transportation-related activity scheduling is becoming more complex due to the growing number of potential locations and extensive opportunities to visit various places. Throughout the years, in the field of transportation several attempts were made to optimize travelers’ activity chains with different parameters to set, but there is a lack of comprehensive solutions. In this research, the activity chain optimization algorithm is applied, which requires high computational efforts. To provide an adequate calibration of the parameters, a sensitivity analysis is conducted. The aim of the analysis is to reveal how changes in the attribute values modify the final outcomes. The relevant parameters, activity chains, transport modes, optimization algorithms, and fitness functions, are identified and considered. For each parameter, an investigation is conducted to reveal its behavior throughout the runs. For example, changes in the population size and crossover function lead to more reliable results, while alteration in the number of generations and the mutation function have no effects on the outcomes. The analysis presents a peculiar behavior of the parameters related to the activity chains. The results can be useful for transportation planners and service providers in the adaptation of the existing network and transportation services to the travelers’ mobility patterns.