tommonks01/swast-benchmark repository overview

⁠Forecasting the daily demand for emergency medical ambulances in England and Wales: A benchmark model and external validation

• Use this image to reproduce the results from the paper: https://osf.io/a6nu5/⁠
• GitHub Repo: https://github.com/TomMonks/swast-benchmarking⁠

⁠Steps to use the Docker image

You will need docker installed on your local machine from https://www.docker.com/products/docker-desktop⁠

• linux users can install: sudo apt-get install docker.io
• Mac OS users can install docker desktop: https://docs.docker.com/docker-for-mac/install/⁠
• Windows user can install docker desktop via WSL2: https://docs.docker.com/docker-for-windows/install/⁠

⁠Pull the image from dockerhub

• A pre-built docker image containing the code and data is available from dockerhub.
• https://hub.docker.com/r/tommonks01/swast-benchmark/⁠

docker pull tommonks01/swast-benchmark:latest

⁠Running the image

# launch the image in interactive mode; map local port 8080->80 on image
docker run -it -p 8080:80 --name swast_docker swast-benchmark

Your terminal will now be in the local docker image. Issue the following commands

# switches to the ambo_benchmark virtual environment
conda activate ambo_benchmark 

# launch jupyter lab. Defaults are no-browser; port=80; ip=0.0.0.0
jupyter-lab

Open the browser on your local machine enter the following address

• localhost:8080 (Juypter lab opens in the repo's directory)

⁠Quickly recreate the tables, charts and figures in the paper:

The results files are included in the repo. To quickly reproduce the tables, figures and statistics included in the results section of the paper run the top level notebook results_summary_main.ipynb. The notebook will display 'most' of the outputs; the appendix tables are not practical to output. All outputs including the appendix can also be found in appendix/

⁠How to conduct a full reproduction run.

Run the following notebooks (the order does not matter)

• ./analysis/model_selection/stage1/00_batch_run_stage1.ipynb
• ./analysis/model_selection/stage2/00_batch_run_stage2.ipynb
• ./analysis/benchmark/00_ensemble-tscv-parallel.ipynb
• ./results_summary_main.ipynb