This project explores ride-hailing dynamics in Chicago using historical taxi trip data as a proxy, enriched with weather and optional point-of-interest data. We cleaned and discretized the data spatially (using H3 hexagons) and temporally, analyzed demand patterns, and developed an interactive KPI-Dashboard. We applied clustering techniques to identify spatial hotspots, built predictive models using support vector machines and neural networks, and compared their performance across different resolutions. Finally, we implemented a reinforcement learning agent to optimize smart charging strategies for an electric vehicle fleet.
- Install Python 3.11
- Install poetry
- Check whether poetry was correctly installed and added to your path.
Use poetry version >= 1.5.0
poetry --version
- Download the Poetry virtual environment with the Python libraries defined in the pyproject.toml file.
Navigate into the "AAA_Team_Assignment_2025" folder and excecute "poetry install".
cd AAA_Team_Assignment_2025poetry install
The Python Notebooks are located in the AAA_Team_Assignment_2025/src directory and should be executed in the order of their numbering, i.e.:
- 00_0_load_data.ipynb
- 01_0_data_preparation.ipynb -> to validate generated samples use: 01_2_data_validity.ipynb
- 02_descriptive_spatial_analytics.ipynb
- 01_1_data_preparation.ipynb
- 03_predictive_analytics.ipynb
- 04_smart_charging_using_reinforcement_learning.ipynb
The additional interactive KPI-Dashboard should be started after the 02_descriptive_spatial_analytics.ipynb notebook, since the resulting data of this notebook is neccessary to run it. The KPI-Dashboard can be executed by the following command in the terminal:
poetry run streamlit run dashboard.pyWhile most of the functionalities of the dashboard should be self explaining here is a list of key usage possibilities:
- Select H3 resolution, time bin, metric, and pickup/dropoff view from the sidebar.
- Explore ride-hailing activity across Chicago via an interactive hex map.
- Click on a hex to view detailed stats and top connected locations.
- Switch between citywide and selected-hex views in the tabs.