R&D Project Structure Musts

[hmehryar]

Todos

• What are the most necessary steps for the project structure and maintainable developments of it

[hmehryar]

Watched and implemented this, it was useful:
https://youtu.be/CAeWjoP525M?si=k-iceKTo34cPQ4g0

[hmehryar]

Project Structure

• Identify The Problem: Jet event paramters (energy loss, $alpha_s$, and $Q_0$) values classification

• Have A Plan

  • Implmenting/training/testing machine learning (ML) and deep learning models until finding successfull classifier models with high fidelity
  • Technology Stack: Python, Tensorflow, Colab, Grib, GPU, Kaggle, [Todo: Hadoop File System]

• Structure Your Directories
  Including one directory for models, and one directory for each parameters (jet_parameters-> [param_name])

• Use Version Control: Github

• Modularize and Componentize Your Code: Having short, reusable, and easy to understand componets

• Documentation: A directory in the project root for general documentations, having following documents will helps

  • Installiation guideline
  • Troubleshooting guideline
  • Contribution guideline
  • Issue template

• Testing

  • Unit Testing
  • Integration Testing

• Dependency Management

  • pip

• CI & CD: Google Test (If applicable), Github actions, [A website that shows the classifier resutls (maybe in annimation)]

• Code Review

[hmehryar]

another useful one
https://youtu.be/MaIfDPuSlw8?si=-flYixkx1QzTGBnW

[hmehryar]

for now, it's a good idea to go with cookiecutter structure and choose what's the best for the project meanwhile

[hmehryar]

having a preprocess_dataset.py and model_validation.py can be useful for each parameter classification folder, I'm trying to practice this to see how practical it is for our case

[hmehryar]

adding an scirpt _ruuner directory can help to keep the code base clean