Welcome to FromPlateToHeart, my inaugural data science project aimed at promoting heart health through personalized product recommendations. This repository captures my exploration, insights, and methodologies, all centered around the extensive Open Food Facts dataset.
As my first project for my Machine Learning diploma, FromPlateToHeart signifies my commitment to utilizing data science to empower individuals with heart-healthy dietary recommendations. Leveraging the Open Food Facts dataset, I've created an application that delivers personalized suggestions aligned with cardiovascular wellness.
In this notebook, I meticulously prepared the Open Food Facts dataset for analysis:
- First Filtering of Interest: Focused on products with names and sold in France 🇫🇷.
- Missing Values Thresholds: Managed missing values with defined thresholds.
- Aberrant Value Handling: Addressed aberrant values effectively.
- Atypical Values Handling: Employed the interquartile range technique to enhanced data integrity.
This notebook dwells on :
- Univariate Analysis: Explored nutrition score and grade distributions.
- Bivariate Analysis: Examined fat content vs. nutrition grade.
- Multivariate Analysis: Explored complex interactions for informed insights.
- Application Output: Applied findings to enhance the recommendation engine.
For deeper insights into my project, check out my presentation slides within the repository (in French).