search

gjuneja0080 / SHProject

2 stars 0 forks source link

Code organisation #3

Open olexandr-konovalov opened 4 years ago

olexandr-konovalov commented 4 years ago

I have seen Jupyter notebooks in the root directory of the project - GitHub usefully renders them, so I don't even have to open them locally.

And I would not be able to run them locally without changes, because they contain local paths, e.g. /Users/gopaljuneja/Desktop/Reproduced_MEK/BDJ_Dandora_data.csv which do not exist on my machine. That's not following good reproducibility practices. I suggest to put data files in this repository (in one or more subdirectories, e.g. data, original etc.) and use local paths, assuming that notebooks will be in the root directory of the project.

I think you should merge PR #1 and add missing files (or redo it by direct commits to the master branch) to ensure that all is kept in one place.

Furthermore, you can then consider placing parts of your project in a public repository, making your Jupyter notebooks runnable on Binder and possibly on Azure Notebooks, using the template I made at https://github.com/rse-standrewscs/python-binder-template. Nicely, the authors of the original paper made their code & data available under CC 4.0 license, so you may include their data files there with appropriate attribution.

Also, you may find Jupyter notebooks from some of my CS2006 Python lectures useful: https://studres.cs.st-andrews.ac.uk/CS2006/Lectures/Python/ (in particular, lectures 7 and 8).

olexandr-konovalov commented 4 years ago

Also, needed some meaningful README, with instructions how to run and brief descriptions of files and directories.