To build the model, use the following command (inside the repository folder):
docker build -t <image_name> .
After building the image, run the model using docker run
. Utilize the --model_type
flag to choose the desired model type. Currently, the available models are: "geographical", "feature", or "mixture".
Running the model involves executing MCMC sampling and then performing predictions in a single run within the Docker container. To save the generated objects from the run, utilize a volume or a bind mount.
For instance, to run the mixture model using a bind mount, execute the following command:
docker run <image_name> -v <path/in/host>:/app/data <image_name> --model_type mixture