跳到主要内容

A Complete Workflow Guide: Using VS Code, Miniconda, and Git for Research Projects

· 阅读需 4 分钟
Liangchao Deng
Liangchao Deng
Ph.D. Candidate @ SHZU @CAS-Cemps

Project Overview​

This playbook outlines a reproducible workflow for Python-focused research—ideal for data analysis, remote sensing, crop modeling, and computational plant science. Follow the sections in order or jump to the one you need.

A Complete Workflow Guide: Using VS Code, Miniconda, and Git for Research Projects

1. Environment Setup​

1.1 Install Visual Studio Code​

  • Download the installer from code.visualstudio.com.
  • Recommended extensions: Python (Microsoft), Jupyter, GitLens, Pylance, Markdown All in One, Remote - SSH.

1.2 Install Miniconda​

conda --version
  • Keep Conda updated:
conda update conda

1.3 Install Git​

git --version
git config --global user.name "Your Name"
git config --global user.email "you@example.com"

2. Project Initialization and Environment Management​

2.1 Create the Project Folder​

mkdir cotton_modeling
cd cotton_modeling

2.2 Initialize Git​

git init

2.3 Create the Conda Environment​

conda create -n cotton python=3.10
conda activate cotton

2.4 Install Core Packages​

conda install numpy pandas matplotlib scikit-learn
conda install -c conda-forge opencv open3d

2.5 Share the Environment​

conda env export > environment.yml

To reproduce the environment elsewhere:

conda env create -f environment.yml

3. Set Up the Project in VS Code​

3.1 Open the Workspace​

  • Launch VS Code → File > Open Folder… → choose the project directory.
  • Select the Python interpreter (Command Palette > Python: Select Interpreter) and pick conda: cotton.

3.2 Suggested Project Structure​

cotton_modeling/
├── data/              # Raw & processed datasets (not committed)
├── notebooks/         # Jupyter notebooks
├── scripts/           # Core Python modules
│   ├── preprocessing.py
│   ├── modeling.py
│   └── visualization.py
├── results/           # Generated figures, tables, reports
├── environment.yml    # Conda spec for reproducibility
├── README.md          # Project overview and usage
└── .gitignore

3.3 .gitignore Essentials​

__pycache__/
*.pyc
*.ipynb_checkpoints
data/
results/
.env

4. Git Workflow (Single Researcher)​

4.1 Stage and Commit Changes​

git add .
git commit -m "Initial commit: data preprocessing pipeline"

4.2 Connect to GitHub​

git remote add origin https://github.com/yourname/cotton_modeling.git
git branch -M main
git push -u origin main

4.3 Sync Regularly​

git pull origin main
git push origin main

4.4 Use Feature Branches​

git checkout -b feature-light-simulation
# Implement changes...
git add .
git commit -m "Add light simulation module"
git push origin feature-light-simulation

5. Collaboration Workflow​

  1. Fork the repository and clone locally:

    git clone https://github.com/leader/cotton_modeling.git

  2. Create a feature branch:

    git checkout -b analysis-update

  3. Commit and push updates:

    git add .
    git commit -m "Update canopy reflectance model"
    git push origin analysis-update

  4. Open a pull request on GitHub for review and merging.

6. Maintenance and Reproducibility​

  • Keep environments current: conda env export > environment.yml
  • Document clearly: Maintain README.md with project overview, requirements, usage, and data notes; use docstrings for modules.
  • Tag releases: git tag -a v1.0 -m "First release" then git push origin v1.0.
  • Manage data responsibly: Keep raw data read-only, avoid committing large binaries, update .gitignore to exclude generated files.

7. Typical Research Project Flow​

  1. Initialize the repository with Git.
  2. Create and activate the Conda environment.
  3. Develop scripts and notebooks in VS Code.
  4. Commit frequently and push to GitHub.
  5. Branch for experiments or new modules.
  6. Export results and environment descriptors.
  7. Reference commit hashes or tags in publications for transparency.

8. Common Issues and Fixes​

IssueQuick Fix
VS Code cannot find the Conda environmentUse Python: Select Interpreter and choose the correct Conda env.
git push authentication errorsRefresh your GitHub token or sign in again using gh auth login.
Conda dependency conflictsRun conda clean --all or recreate the environment from environment.yml.
Jupyter kernel missingInstall kernel: python -m ipykernel install --user --name=cotton.

9. Final Notes​

Adopting VS Code, Miniconda, and Git as a unified workflow delivers:

  • Reproducibility: every environment and code change is versioned.
  • Transparency: collaboration and provenance are traceable.
  • Efficiency: tooling accelerates experimentation and debugging.

Author: Liangchao Deng
Website: smiler488.github.io

CN Community

EN Community

More