Summary
Developed an end-to-end modeling framework and scenario engine to support planning decisions across the shea supply chain, including a digital twin of origin procurement, processing schedules, and product valorization. Delivered actionable playbooks and a prototype tool in Dataiku to standardize inputs/outputs and enable repeatable scenario analysis.
Context & Objectives
- Map upstream kernel sourcing, processing capacities, and demand nodes to quantify bottlenecks and cost drivers.
- Enable “what-if” scenario testing (origin mix, processing schedules, early kernel buying).
- Translate insights into planning guidelines used by procurement and operations.
Approach
- Unified data model spanning origin → processing → output qualities; scenario parameters encapsulated in a configuration layer.
- Multi-level objectives (cost, service, valorization) with constraints (capacity, quality windows, logistics).
- Prototype scenario tool in Dataiku for standardized inputs/outputs and reproducibility.
- Visual analytics to compare baseline vs. alternative strategies.
Impact
- Identified strategies with potential savings of €1M+ per year (e.g., early kernel buying, origin optimization, improved processing schedules).
- Improved cross-functional alignment via shared digital twin and planning playbooks.
- Reduced analysis cycle time by standardizing scenario computation.