Food Product Data Analysis for Nutrition App

Project Overview

This project explores a public dataset of food products to develop a scoring system for a nutrition recommendation application. The app evaluates products based on three key metrics: health score, environmental score, and packaging impact. The project involved cleaning and analyzing data, designing scoring methods, and creating insights to assist users in making informed food choices.

Objectives

• Clean and preprocess food product data to address missing values and inconsistencies.
• Analyze data to derive insights for nutritional, environmental, and packaging scores.
• Implement scoring systems to assess health, ecological, and packaging impacts.
• Demonstrate the feasibility of a nutrition recommendation app with computed scores.

Tools & Techniques

• Python Libraries: Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn
• Data Processing: Missing value imputation using KNN, statistical analysis, and data normalization.
• Analysis:
  • Univariate: Distribution histograms and normality tests.
  • Bivariate: Correlation analysis and scatter plots.
  • Multivariate: Principal Component Analysis (PCA) for dimensionality reduction.

Key Insights

• Health Score:
  • Categories like vegetables, rice, and milk have the highest scores.
  • Increased additives and palm oil content reduce the health score.
• Environmental Score:
  • Top countries: Poland, Sweden, and France.
  • Environmental score decreases with higher packaging impact and shipping distance.
• Packaging Score:
  • Fresh products score higher, while plastic-heavy packaging scores lower.
  • Certifications like FSC and Green Dot are positive indicators.

Deliverables

• Jupyter Notebooks:
  • Data Cleaning: Handling missing values and anomalies.
  • Data Exploration: Statistical and visualization analysis.
• Presentation:
  • Summarizes key findings and recommendations.
• Proposed App:
  • Scores products on health, environment, and packaging for informed decisions.

Future Improvements

• Include detailed packaging and shipping data for precise ecological scoring.
• Enhance prediction models with advanced machine learning techniques.
• Expand analysis to integrate consumer feedback for score validation.

Why This Project?

This project showcases expertise in:

• Data cleaning, analysis, and visualization.
• Machine learning techniques for imputation and prediction.
• Application of data science to real-world consumer problems.