Agent-based Activity Generation of Runners for City Infrastructure Planning

In this project, an agent-based model is developed by using MESA framework in Python 3+ to generate the daily activities for a population of people. Specifically, focusing on runners and generating a set of routes that typical runners would take on their daily runs. A heat map is automatically created at the end of the simulation to compare with the Strava heat map.

Features

Simple virtual world

• Create a simple grid world to represent a simple world which has normal roads, trails, forests, houses, and grass.
• Initialize a number of runners to roam around the map and follow the state machines of its corresponding behaviors.

Virtual city

• Transform an actual street network dowloaded by OSMnx package of a neighborhood in Portland city into a grid world.
• Encode several real behaviors of runners into this agent-based model and simulate it with 500 agents on the grid world.
• Generate a heat map at the end to identify roads that are used by runners and the level of being used.

Framework/Package used in Python

• MESA
• OSMnx
• NetworkX
• Pandas
• Numpy
• Matplotlib
• Seaborn

Application

This model is an example to study about the runners of a neighborhood in Portland city. However, this is transferable and could be used for other areas as all street networks data from OpenStreetMap are available and could be downloaded by using the OSMnx package.