This project is part of the Applied Data Science program, integrating descriptive, predictive, and prescriptive analytics into a unified supply chain solution. It solves a real-world transportation problem by minimizing delivery costs through mathematical optimization.
The application follows a rigorous three-phase industrial workflow:
- Descriptive Phase: Visualizes the logistics network (warehouses and customer nodes) using geographic mapping and inventory metrics.
- Predictive Phase: Implements a Linear Regression engine to forecast demand for the next business day based on historical data.
- Prescriptive Phase: Utilizes Linear Programming (PuLP) to solve the cost-minimization objective, determining the optimal allocation of units from hubs to customers.
- Optimization: PuLP (Linear Programming).
- Predictive Modeling: Scikit-Learn (Linear Regression).
- Frontend: Streamlit (Executive Dashboard).
- Data Management: SQLite & Pandas.
- Validation: Pydantic for schema enforcement.
- Automated Data Seeding: Includes a
seed_data.pyscript to initialize the logistics environment. - Mathematical Precision: The optimizer considers warehouse capacity constraints and customer demand requirements.
- Professional Engineering: Full type-hinting, modular logic, and centralized configuration management.
Q: Why use Linear Programming instead of a simple Greedy Algorithm? A: A greedy algorithm might find a local optimum by choosing the cheapest route for one customer at a time, but it often fails to minimize the global cost. Linear Programming evaluates all constraints simultaneously to find the absolute mathematical minimum.
Q: How does the system handle a situation where demand exceeds capacity?
A: The route_optimizer.py script detects "Infeasible" status. In a real-world scenario, this triggers a business alert to prioritize high-value customers or increase warehouse shifts.
Q: How is the objective function defined in this project? A: The objective function is the minimization of the sum of (Units Shipped × Euclidean Distance) across all active routes, subject to supply and demand constraints.
📄 License This project is distributed under the MIT license. Its purpose is strictly educational and research-based, developed as an Applied Data Science solution.
Note for recruiters: This Capstone project demonstrates the ability to translate complex business constraints into mathematical models. It showcases mastery over the entire data lifecycle—from raw SQL data to a prescriptive engine that provides actionable strategic decisions.
Autor: JUAN S. Contacto: https://github.com/johnyse99