📦 E-Commerce Warehouse Optimization

Interactive Analytics Platform for Supply Chain Optimization using Linear Programming

🚀 Overview

A comprehensive Linear Programming solution for multi-warehouse inventory allocation that achieves 99%+ order fulfillment and $44.9M profit improvement through optimal resource distribution and cost minimization.

This project optimizes inventory allocation across 6 warehouses serving 10 delivery regions with 300+ products, solving the complex problem of balancing transportation costs, holding costs, and service level targets.

---

🎯 Key Results

Metric	Before Optimization	After Optimization	Improvement
Order Fulfillment	45%	99%+	+54 pp
Stockout Rate	55%	<5%	-50 pp
Annual Profit	-$9.5M (loss)	+$35.4M	$44.9M
System Cost	$36M	Optimized	Minimized

Key Achievements:

• ✅ 99% fulfillment rate (up from 45%)
• ✅ $44.9M profit improvement from optimization
• ✅ $3.7M savings opportunity identified (20% transport cost reduction)
• ✅ Optimal allocation across 6-warehouse network

---

🛠️ Features

Interactive Streamlit Dashboard

• 🏠 Executive Summary: High-level KPIs and business impact metrics
• 📈 Performance Analysis: Deep dive into fulfillment, delivery, warehouse, and transport metrics
• 🔄 Scenario Comparison: Sensitivity analysis across 9 optimization strategies
• 🗺️ Network Visualization: Interactive global map of warehouse-region connections
• 💡 Strategic Insights: Data-driven recommendations with ROI projections
• 🔬 Technical Documentation: Mathematical formulations and methodology
• 📊 Validation Logs: Complete audit trail of calculations and assumptions

Optimization Engine

• Multi-Commodity Network Flow optimization
• Linear Programming using PuLP/CBC solver
• Flow Capacity Model (12× inventory turnover)
• Cost Minimization (transport + holding + stockout penalties)
• Constraint Optimization (capacity, demand, service levels)

---

💻 Technology Stack

• Python 3.10+ - Core programming language
• PuLP - Linear Programming and optimization
• Pandas - Data manipulation and analysis
• Streamlit - Interactive dashboard framework
• Plotly - Advanced data visualization
• NumPy - Numerical computing

---

📁 Project Structure

Ecommerce_Warehouse_Optimization/
│
├── streamlit_dashboard.py    # Main dashboard application
├── requirements.txt          # Python dependencies
├── .devcontainer/            # Development container config
│
├── results/                  # Generated optimization results
│   ├── scenario_comparison_kpis.csv
│   ├── cost_breakdown_comparison.csv
│   ├── network_nodes.csv
│   ├── network_edges.csv
│   └── Baseline/             # Baseline scenario results
│       ├── shipments.csv
│       ├── stocking.csv
│       ├── stockouts.csv
│       └── kpis.json
│
└── README.md                 # Project documentation

---

⚙️ Installation & Setup

Prerequisites

• Python 3.10 or higher
• pip package manager

Step 1: Clone the Repository

git clone https://github.com/om-gorakhia/Ecommerce_Warehouse_Optimization.git
cd Ecommerce_Warehouse_Optimization

Step 2: Install Dependencies

pip install -r requirements.txt

Required Packages:

streamlit>=1.28.0
pandas>=2.0.0
numpy>=1.24.0
plotly>=5.17.0
pulp>=2.7.0
json
pathlib

Step 3: Run the Dashboard

streamlit run streamlit_dashboard.py

The dashboard will open in your default browser at http://localhost:8501

---

📊 How It Works

Problem Statement

Challenge: An e-commerce company has:

• 6 warehouses with limited capacity
• 10 delivery regions with varying demand
• 300+ products to allocate
• 45% fulfillment rate (55% stockouts)
• Negative profitability

Goal: Maximize order fulfillment while minimizing total system costs

Solution Approach

1. Flow Capacity Model

Instead of static inventory, we model continuous replenishment:

Flow Capacity = Static Inventory × Annual Turnover Rate
              = 50,000 units × 12 turns/year
              = 600,000 units/year capacity

2. Objective Function

Minimize total system cost:

Minimize: Transportation Cost + Holding Cost + Stockout Penalties

3. Key Constraints

• Demand Satisfaction: All demand must be fulfilled or recorded as stockout
• Capacity Limits: Shipments cannot exceed warehouse capacity
• Service Targets: 95% fulfillment, 3-day delivery window
• Non-negativity: All quantities must be ≥ 0

4. Optimization

Using Linear Programming (PuLP with CBC solver) to find the mathematically optimal allocation.

---

💼 Business Impact

Problems Solved

Problem	Solution	Result
Low Fulfillment (45%)	Flow Capacity Model + Optimal Allocation	99%+ fulfillment
High Stockouts (55%)	Linear Programming optimization	<5% stockouts
Unprofitable Operations	Cost minimization with 10× CLV penalty	$44.9M profit gain
Inefficient Transport	Carrier negotiation opportunity	$3.7M savings potential

Strategic Recommendations

🔴 Priority 1: Transportation Cost Negotiation (6 months)

• Expected savings: $3.0-3.7M annually
• Action: Negotiate 15-20% rate reduction with carriers
• ROI: Highest sensitivity

🟡 Priority 2: On-Time Delivery Improvement (18 months)

• Target: 95%+ on-time rate
• Action: Add 2-3 regional distribution centers
• ROI: High

🟢 Priority 3: Capacity Monitoring (Ongoing)

• Current: 68% average utilization
• Headroom: 25% volume growth before expansion
• Action: Quarterly monitoring

---

🔬 Technical Details

Mathematical Formulation

Decision Variables:

• x[i,j,p] = Quantity of product p shipped from warehouse i to region j
• y[i,p] = Binary indicator (1 if product p stocked at warehouse i)
• s[j,p] = Stockout quantity for product p in region j

Objective Function:

Minimize Z = ΣΣΣ c[i,j] × x[i,j,p]           (Transportation)
           + ΣΣ h[i] × inventory[i,p]      (Holding)
           + ΣΣ penalty[p] × s[j,p]      (Stockouts)

Subject to:

• Demand satisfaction: Σ x[i,j,p] + s[j,p] = demand[j,p]
• Capacity constraint: Σ x[i,j,p] ≤ flow_capacity[i,p]
• Service level: On-time delivery ≥ 95%

Performance

• Solve Time: ~2 seconds
• Problem Size: ~14,000 variables, ~5,000 constraints
• Optimization Status: Optimal solution guaranteed
• Algorithm: Branch-and-Cut with CBC solver

---

📝 Use Cases

This solution is applicable to:

• ✅ E-commerce fulfillment network optimization
• ✅ Multi-warehouse inventory allocation
• ✅ Supply chain cost reduction initiatives
• ✅ Service level improvement projects
• ✅ Capacity planning and expansion analysis
• ✅ Warehouse network design
• ✅ Transportation optimization

---

📦 Scenarios Analyzed

The project evaluates 9 optimization scenarios:

1. Baseline - Current optimal allocation 2-4. Capacity Expansion - 10%, 20%, 30% increase 5-6. Transport Cost Reduction - 10%, 20% decrease
2. Transport Cost Increase - 10% increase 8-9. Higher Service Targets - 97%, 99% fulfillment

---

👥 Team

DBA5103 Operations Research & Analytics
National University of Singapore (NUS)

• Om Gorakhia
• Prisha Shah
• Nourah Algiffari
• Pranay Samineni
• Sanya Rajpal

---

📌 Next Steps

To extend this project:

1. Real-time Integration: Connect to live inventory systems
2. Demand Forecasting: Add ML-based demand prediction
3. Dynamic Pricing: Incorporate pricing optimization
4. Route Optimization: Detailed last-mile delivery routing
5. Multi-period Planning: Extend to monthly/quarterly planning horizon
6. Sustainability Metrics: Add carbon footprint optimization

---

📝 License

This project is part of academic coursework at NUS Business School.

---

🚀 Quick Start

# Clone repository
git clone https://github.com/om-gorakhia/Ecommerce_Warehouse_Optimization.git

# Navigate to directory
cd Ecommerce_Warehouse_Optimization

# Install dependencies
pip install -r requirements.txt

# Launch dashboard
streamlit run streamlit_dashboard.py

Dashboard URL: http://localhost:8501

---

❓ FAQ

Q: What data is required to run this optimization?
A: The model requires warehouse capacities, product dimensions, demand by region, transportation costs, and service level targets.

Q: Can this work for my business?
A: Yes! The framework is adaptable to any multi-location inventory allocation problem. Adjust parameters in the optimization engine to match your specific requirements.

Q: How long does optimization take?
A: The CBC solver finds optimal solutions in ~2 seconds for networks of this size (6 warehouses, 10 regions, 300 products).

Q: What if I don't have optimization results?
A: Run the analysis engine first (check code comments for execution). Pre-generated results are included in the /results folder.

---

📞 Contact

For questions or collaboration:

• Email: om.gorakhia@u.nus.edu
• LinkedIn: linkedin.com/in/omgorakhia
• GitHub: @om-gorakhia

---

⭐ If you find this project helpful, please give it a star!

---

Built with ❤️ for supply chain optimization | NUS Business Analytics 2025