This repository contains materials for the SENSE practical on agent-based modelling as part of the training week on Machine Learning and AI. The materials for the practical are separated into 4 subdirectories.

The ./data/ subdirectory contains the data required for the practical, as well providing a location in which to place any data outputs that may be produced.

The ./notebooks/ subdirectory contains the jupyter notebooks that will be run for the practical. The practical consists of 4 notebooks. The first of these notebooks provides an introduction to the practical, giving an overview of agent-based models, the pedestrian agent-based model used in this practical and how it will be used in the practical. The second notebook briefly outlines how we go about running an instance of the agent-based model with user-specified values for parameters. The third notebook provides some activities in which we analyse some of the data provided with the goal of identifying appropriate values for various speed parameters in the model. The fourth notebook contains the main activities of the practical in which we use the speed parameter values found in the previous notebook to run the model with the aim of calibrating other agent parameters in the model (specifically, the agent birth-rate).

The ./src/ subdirectory contains any source code used for the practical. This is divided into two purposes:

• Code used for preprocessing data
• Code used for the agent-based model

The ./docs/ subdirectory contains any documentation relevant to the practical.

This section of the README will outline how to get started with the practical. The process of getting started involves the following steps:

• Getting hold of the practical materials
• Installing the necessary python packages
• Getting access to the practical data
• Starting up jupyter-notebook/jupyter-lab

This practical can be approached via two avenues:

• GitHub
• Minerva

The requirements for obtaining the practical from GitHub are:

• git
• make
• python 3.x (preferably Anaconda Python)

The requirements for obtaining the practical from Minerva are that you have access to the SENSE CDT Minerva organisation. The advantage of obtaining the materials via Minerva is ease of access - you will be able to download a zipped folder, unzip it and get started. The advantage of obtaining the materials via GitHub is that you will be able to pull any changes that I make to the practical in close to real-time.

(The remainder of this section will assume that you are using Anaconda Python)

The process of getting started from GitHub consists of the following steps:

• Cloning down the repository with git
• Setting up a python environment with conda
• Setting up the data with python 3, make and tar

The first step in getting started with this practical is to get hold of this repository. We can do this by using git to clone the repository onto our own machines. This is achieved by opening the terminal/command prompt and running the following command:

git clone https://github.com/ksuchak1990/sense_practical

This will download a copy of the repository onto our machines. We then wish to move into the folder that we have just downloaded, which we can do by running the following command:

cd sense_practical

After getting hold of the repository, we wish to set up an environment. This repository contains an environment.yml file, which documents the python packages that are required to run this practical. The easiest way to install this list of packages is to open your terminal or anaconda prompt and run the following command:

conda env create -f environment.yml

This will create a conda environment named sense-abm which will contain all of the required packages. The environment can then be activated by running the following command:

conda activate sense-abm

Be aware that we should ensure that this environment is activated before running any of the notebooks for the practicals.

After having installed all of the required packages under an environment and having activated the environment, we wish to set up the data. At present the input data exists in ./data/inputs as a tar zip (frames.tar.gz). If you have make and tar installed on your computer then the process of setting up the data can be automated by running:

make frames

Once we reach this stage, we are ready to get going with the notebooks in ./notebooks/!

If obtaining this practical from Minerva, you don't need to worry about any of the data processing issues mentioned above - you can just download the zipped folder, set up a Python environment and you're ready to go!

This repository contains an environment.yml file, which documents the python packages that are required to run this practical. The easiest way to install this list of packages is to open your terminal or anaconda prompt and run the following command:

conda env create -f environment.yml

This will create a conda environment named sense-abm which will contain all of the required packages. The environment can then be activated by running the following command:

conda activate sense-abm

Be aware that we should ensure that this environment is activated before running any of the notebooks for the practicals.

• Look at DUST repository
• Contact me: [email protected]