[1L4-OS-4a] OS-4

[1L4-OS-4a-02] Visualization for Improving Parameter Interpretability in Pedestrian Flow Simulations with Multiple Indicators

〇Chisa Mori¹, Mei Takeda², Masaki Onishi³, Takayuki Itoh¹ (1. Ochanomizu University, 2. University of Tsukuba, 3. National Institute of Advanced Industrial Science and Technology)

Keywords:Visualization, social phenomenon

Pedestrian flow simulation is controlled by parameters that determine the guidance of the human flow.The simulation results are typically evaluated by several indicators such as congestion and travel time.It is necessary to interpret the indicators of the simulation to develop a guidance plan suitable for the real world.Meanwhile, there are trade-offs between multiple indicators.It is often challenging to find the equilibrium indicators that bring desired guidance for the simulation users. Furthermore, selecting parameter values that provide optimal guidance is sensitive due to the complexity of the relationship between evaluation indicator values and parameters. Therefore, this study proposes a visualization method to interpret parameters in terms of multiple indicators to reveal the relationship between parameters and evaluation indicators and facilitate the interpretation of pedestrian flow simulations considering the trade-offs.In this study, we visualize the relationship between the parameters and the linearly weighted sum of multiple evaluation values using parallel coordinate plots, which are calculated by linearly weighting the values to accurately represent the balance of evaluation indicators that achieve the guidance method desired by users.