TrajectoryVisual

Our P8 project about visualizing simplified trajectories.

Setup

Run directly or use docker (Or other containerization tool)

1. Install uv: pip install uv
2. Run uv sync to set up the environment
3. Download the dataset:
   • Go to: https://www.kaggle.com/datasets/crailtap/taxi-trajectory
   • Download train.csv (Login required)
   • Extract the files to the datasets folder within the repository
   • Copy the first 101 lines into a csv file called small_train.csv
4. (Optionally get the pickle files)
5. Run uv run main.py to execute

Using Docker

Make sure docker engine is installed: https://docs.docker.com/engine/install/ Prepare the folder by getting the dataset, and optionally the pickle files. Then build the image:

docker build -t trajectory .

Create and run a container:

docker run -dit -e JOB_OUTPUT_DIR=/results/{This runs output directory} --rm --mount type=bind,src=./results/,dst=/results/ trajectory {Command line arguments}

Example:

docker run -dit -e JOB_OUTPUT_DIR=/results/0 --rm --mount type=bind,src=./results/,dst=/results/ trajectory --knn 1 --range 1 --similarity c

The run commands can be generated by running the helper script: generate_grid_search_commands.py

uv run generate_grid_search_commands.py