Datasets are needed to run the notebook. You can either download them from Roma Open Data's site and then run the preprocessing script. Otherwise download a pre-processed ready for use version from the Releases page of this repository.
For running the scripts and the notebook you need a Python 3.9+ installation with Pandas and a container runtime like Docker or Podman.
Since many rows in the accidents don't have a coordinate but just a street name, we need to obtain the coordinates ("geocoding") by running a Nominatim instance. To do so please run the following:
# Run Nominatim
./scripts/run-nominatim.sh
# Do the pre-processing
./scripts/preprocessing.py datasets/raw
From the root directory:
./scripts/run-jupyter.sh