Graph-based modelling of COVID-19 spread in the Caribbean developed for the Centre for Design Informatics at The University of Edinburgh.
Please note the model is intended for educational purposes only.
The repository does not include real Caribbean based data.
Project dependencies are listed in env.yaml file, which can be used to create a virtual environment with conda.
conda env create -f env.yaml
conda activate covid19-caribbean-model
See conda environments documentation for more information.
See the data description for detailed information about the origin and expected formatting of the data files.
Run a simulation for a single set of parameters. The model is run 10 times by default and the results are averaged.
python -m src.run_single -i sample
usage: run_single.py [-h] -i file [-o file] [-n int] [-lg str] [-ng] [-q]
[-gp file] [-dtp file]
Runs the simulation once based on the files provided
optional arguments:
-h, --help show this help message and exit
-i file, --input_spec_file file
JSON file specifying locations of data files required
for the specification. See input files specification
in data/README.md for details. 'real' defaults to
data/real/input-spec.json; 'sample' defaults to
data/sample/sample-input-spec.json
-o file, --output_filename file
Filename where the results should be saved.
data/output/sample-output.json by default.
-n int, --number_of_runs int
Number of runs of the simulation. If bigger than 1,
the simulation results are averaged across the runs.
10 by deafult.
-lg str, --load_graph_dir str
When supplied, attempts to load graph from a file
{community_name}-graph.pkl in specified directory.
-ng, --new_graph Construct a graph on per simulation basis instead of
once per run.
-q, --quiet Do not print results to console.
-gp file, --graph_plot_filename file
When supplied, saves PNG graph visualisation to given
filename.
-dtp file, --doubling_time_plot_filename file
When supplied, saves PNG doubling time plot to given
filename.
Run a simulation for each combination of app inputs provided.
python -m src.run_all -i sample
usage: run_all.py [-h] -i file [-o dir] [-n int] [-tasks int] [-chunk int]
[-cpu int] [-uinter] [-dinter] [-gensample] [-lg str] [-ng]
[-sp]
Runs the simulation for each combination of app inputs provided
optional arguments:
-h, --help show this help message and exit
-i file, --input_spec_file file
JSON file specifying locations of data files required
for the specification. See input files specification
in data/README.md for details. 'real' defaults to
data/real/input-spec.json; 'sample' defaults to
data/sample/sample-input-spec.json
-o dir, --output_dir dir
Directory where the results should be saved.
data/output/ by default.
-n int, --number_of_runs int
Number of runs per single simulation. If bigger than
1, the simulation results are averaged across the
runs. 10 by default.
-tasks int, --max_tasks_per_child int
Number of maximum tasks per child subprocess. 1 by
default.
-chunk int, --chunksize int
Chunksize of multiprocessing pool mapping. 1 by
default.
-cpu int, --cpu int Number of worker processes to use. Number of CPU cores
by default.
-uinter, --use_intermediary
Compute only variants not already present in
intermediary/ directory.
-dinter, --delete_intermediary
Delete the intermediary/ directory after the
simulations finish. intermediary/ directory stores
graphs and results of every variant in separate files
which are joint together after all simulations finish.
-gensample, --generate_sample
Do not run the simulation and return dummy results
instead.
-lg str, --load_graph_dir str
When supplied, attempts to load graph from a file
{community_name}-graph.pkl in specified directory.
-ng, --new_graph Construct a graph on per simulation basis instead of
once per community.
-sp, --single_processing
Do not use multiprocessing.
Graphs can be precomputed and saved to a file so that the simulation can load them and skip the graph building step.
python -m src.generate_graphs -i sample -c A,B
usage: generate_graphs.py [-h] -i file [-o dir] [-c csv names]
Generate and save graphs for communities
optional arguments:
-h, --help show this help message and exit
-i file, --input_spec_file file
JSON file specifying locations of data files required
for the specification. See input files specification
in data/README.md for details. 'real' defaults to
data/real/input-spec.json; 'sample' defaults to
data/sample/sample-input-spec.json
-o dir, --output_dir dir
Directory where the results should be saved.
data/output/tmp by default.
-c csv names, --communities csv names
Comma-separated list of communities for which to
generate graphs. If not supplied, generates graphs for
all communities listed in the community data file.
A suit of unit tests is provided and can be run using
python -m pytest tests/