Nutrition & Obesity Trends Analysis (Bell-Labs)
Bell Labs (repository name) implements a reproducible analytics stack for global diet and health outcomes. Raw FAO nutrition and population inputs and WHO obesity prevalence are cleaned, harmonized, and joined through a staged pipeline (`run_pipeline.py`): FAO preprocessing, obesity standardization, food-group mapping, panel construction, and a final master merge producing `master_panel_final.csv`. The dataset supports macro-level questions on per-capita energy, protein, fat, food-group shares, and obesity trends. Analysis layers include scripted EDA (`perform_eda.py`, `extended_eda.py`), interactive Plotly dashboards (`interactive_plot.py`), and notebook-driven exploration with documented methodology, data dictionary, and research notes under `doc/`.
Timeline
Multi-month
Role
Data / ML Engineer
Team
Solo
Status
Technology Stack
Key Features
Key Learnings
- Panel data construction for international health and agriculture statistics
- Python packaging of multi-step ETL with clear folder conventions (raw → cleaned → panels → final)
- Bridging notebooks and scripts for both exploration and repeatable runs
- Communicating nutrition–obesity relationships responsibly with documented limitations
Key Challenges
- Aligning country names and codes across FAO and WHO sources
- Managing missing values and interpolation policy without overstating certainty
- Keeping intermediate artifacts organized so the pipeline is rerunnable and auditable
- Explaining high-dimensional nutrition structure to non-technical readers via clear visuals