Decision Intelligence System integrating Machine Learning and Linear Optimization (Pyomo + CBC) to minimize freight cost and delays across 37 logistics hubs in the Northeastern U.S.
This project builds a Decision Intelligence System that integrates machine learning and optimization to improve freight network performance.
It simulates 180k+ shipments across 37 logistics hubs, predicts operational metrics, and optimizes routing decisions using Pyomo + CBC.
1. Data Intelligence (EDA)
- Cleaned and visualized operational data.
- Identified correlations (Cost–Distance: 0.77, Delay independence, etc.).
2. Predictive Intelligence (XGBoost)
- Predicted shipment cost (R² ≈ 0.78)
- Predicted delay probability (R² ≈ 0.31)
- Predicted fuel consumption (R² ≈ 0.99)
3. Prescriptive Intelligence (Pyomo + CBC)
- MILP model minimizing
0.7 × Cost + 0.3 × Delay - Service-level ≥95%, delay ≤3 hrs
- 13% total composite cost reduction
| Metric | Baseline | Optimized | Improvement |
|---|---|---|---|
| Composite Cost (Cost + Delay Weighted) | 572,817 | 498,886 | 12.9% ↓ |
Python · Pandas · Seaborn · Scikit-Learn · XGBoost · Pyomo · CBC Solver
| File | Description |
|---|---|
Intelligent_Freight_Network_Optimization.ipynb |
Full analysis + optimization notebook |
data/northeast_freight_network_raw.csv |
Dataset |
requirements.txt |
Environment dependencies |
The system demonstrates how predictive analytics and linear optimization can jointly drive data-driven decisions in logistics, reducing operational cost while maintaining service quality.
Damodar Satyadeva Madhukar Naraparaju
Master’s in Business Analytics – Saint Peter’s University, NJ
📧 dsd.madhukar@outlook.com