This is a submission for the GitHub Copilot CLI Challenge
What I Built
I built a small Python project that analyzes coffee data from the Coffee Quality Institute (CQI) dataset. The CQI dataset contains detailed scoring for coffee samples from different countries of origin, based on attributes like aroma, flavor, acidity, sweetness, and more.
I wanted to answer simple but fun questions like:
- Which countries produce the highest-rated coffee?
- Does country of origin affect sweetness or acidity?
- What actually separates a top-scoring coffee from an average one?
It’s basically a data-driven way to understand what makes great coffee great. And maybe where to find it 🙂
Demo
Here’s what the data revealed:
Dataset Overview
Top 5 Countries by Total Coffee Score
The highest-ranked countries based on the combined scores of all coffee attributes, showing which origins consistently produce exceptional coffee.
Coffee Profile Comparison by Total Score
Radar chart showing the highest, mid-range, and lowest scoring coffees.
Sweetness by Country
A quick comparison highlighting which countries produce the sweetest and least sweet coffees.
Check out the full analysis here
My Experience with GitHub Copilot CLI
GitHub Copilot CLI genuinely changed how I approached this project. Basically, Copilot was helping with every step: project setup, building, debugging and refactoring.
Project Setup in Seconds
With a single prompt, Copilot scaffolded the entire project:
Create a Python project for data analysis with pandas.
1. Create project directory called 'data-analysis'.
2. Set up a Python virtual environment inside it.
3. Install pandas, flake8 and black in the virtual environment.
4. Configure flake8 and black for code quality.
5. Create README.md.
6. Add a .gitignore file.
Fast Data Exploration
With a simple prompt, I could quickly test hypotheses or look for correlations, for example:
"Analyse correlation between origin country and coffee sweetness. Use python, pandas, matplotlib to visualise the data."
Copilot generated:
- A clean script
- Statistical calculations
- A ready-to-run visualization
Time saved: ~30-45 minutes of writing boilerplate, debugging and tweaking plots. I just reviewed the output and refined the logic. Huge time saver.
What Made It Different
Unlike other AI coding tools I've tried, Copilot CLI:
- Remembers context across commands in the same session (like excluding countries with too few samples)
- Understands project structure in the same session (it knew where to put scripts, how to organize outputs, etc.)
- Suggests meaningful commit messages after changes and runs them when needed
For data analysis projects where you're constantly exploring patterns and experimenting with different visualizations, this workflow is excellent 😊
After seeing these coffee scores, I’m really curious — which country’s coffee is your favorite? ☕
Top comments (1)
very cool project, marina! 💯
also nice to hear about your experience with Copilot CLI. tell us more!
personally, I have a fondness coffee from Colombia and Guatemala ☕