panford/bitrade-graphs

A graph-based approach to modelling bilateral trade between countries

• 👋 Hi, I’m Kobby Panford-Quainoo
• 👀 I’m interested in Machine Learning Research & Engineering
• 🌱 I’m currently working on some ideas pertaining to Deep Generative Models and Climate Simulations
• 💞️ I’m looking to collaborate on any Machine Learning projects
• 📫 I'm reachable by email - [email protected]

bitrade-graphs's Issues

add a license

You can use https://choosealicense.com to decide on a license. I usually use MIT for research code.

Add paper link to repo description

Add https://doi.org/10.13140/RG.2.2.16746.47047 as a website, as shown bellow.

add a README

The README should:

• briefly explain the goal of the project,
• contain instructions to reproduce the work (both to install and run),
• explain what is the content of the repository (for example: what are the notebooks about, what is the data, etc.).

You can look at this example.

don't version temporary files

The __pycache__ folder shouldn't be versioned. It contains temporary files created by the python runtime. Those will be recreated when the code is run again.

You can tell git to ignore files with a .gitignore file. A simple default is:

*~
.DS_Store

# Byte-compiled / optimized files
__pycache__/
*.py[cod]

# IPython checkpoints
.ipynb_checkpoints/

# Local configuration
/.env
/.python-version

state your dependencies

Add a requirements.txt (pip) or environment.yml (conda) that lists the software to be installed alongside its version. That is to make the environment reproducible. It's necessary for the code not to break once dependencies get upgraded, and for other researchers to reproduce the results.

avoid redundancies

It seems that multiple classes and functions are defined both in notebooks and python files. For example, GCNet in Models2.ipynb, training_nb.ipynb, and models.py. Please define them once (in models.py) and import the definition in the notebooks (from models import GCNet).