Warehouse Inventory Optimizer

An AI-powered warehouse inventory optimization system that uses demand forecasting and linear optimization to recommend optimal stock transfers between warehouses, reducing costs and improving service levels.

Features

• Demand Forecasting: Predict future demand using historical order patterns
• Stock Pressure Index (SPI): Calculate inventory pressure for each warehouse-product combination
• Smart Recommendations: Generate optimal transfer recommendations between warehouses
• What-If Analysis: Simulate different scenarios with interactive sliders
• Interactive Dashboards: Visualize data with Plotly charts and graphs
• Cost Optimization: Minimize total logistics costs while maintaining service levels

Prerequisites

• Python 3.8 or higher
• pip package manager

Installation

1. Clone or download the project

   cd intelligent_warehouse_balancer

2. Install dependencies

   pip install -r requirements.txt

3. Run the application

   streamlit run app.py

4. Open your browser Navigate to http://localhost:8501 to access the application.

Data Structure

Warehouse Inventory CSV (data/warehouse_inventory.csv)

Required columns:

• Warehouse_ID: Unique identifier for each warehouse
• Location: Geographic location (city/region)
• Product_Category: Product category (Electronics, Clothing, etc.)
• Current_Stock_Units: Current inventory level
• Reorder_Level: Minimum stock threshold
• Storage_Cost_per_Unit: Cost per unit for storage (INR)

Orders CSV (data/orders.csv)

Required columns:

• Order_ID: Unique order identifier
• Order_Date: Date of the order (YYYY-MM-DD format)
• Origin: Origin location (should match warehouse locations)
• Product_Category: Product category
• Order_Value_INR: Order value in Indian Rupees

How to Use

1. Data Input

• Option A: Use the provided sample data for testing
• Option B: Upload your own CSV files using the sidebar

2. Dashboard Navigation

The application has 5 main tabs:

Overview

• Key metrics and KPIs
• Total warehouses, SKUs, shortage percentage
• Potential cost savings
• Top risk categories

Stock Dashboard

• Stock Pressure Index (SPI) visualization
• Stock vs Reorder Level heatmap
• Detailed stock analysis table

Demand Forecast

• Historical demand trends
• Predicted demand for next period
• Category-wise demand patterns

Rebalancing

• Transfer recommendations table
• Transfer network visualization
• CSV export functionality

What-If Simulator

• Interactive simulation controls
• Scenario analysis (Best Case, Worst Case, Current)
• Real-time impact visualization

3. Simulation Parameters

Use the sidebar sliders to:

• Demand Change: Simulate demand fluctuations (-50% to +50%)
• Storage Cost Change: Simulate cost variations (-30% to +30%)

Key Algorithms

Stock Pressure Index (SPI)

SPI = (Current_Stock_Units - Reorder_Level) - Predicted_Demand

• Negative SPI: Shortage risk (needs stock)
• Positive SPI: Surplus (can donate stock)
• Zero SPI: Balanced inventory

Transfer Optimization

1. Identify warehouses with SPI < 0 (shortage)
2. Find warehouses with SPI > 0 (surplus) for same product category
3. Calculate optimal transfer units
4. Estimate cost savings

Cost Savings Calculation

Savings = Transfer_Units × (Donor_Storage_Cost - Receiver_Storage_Cost)

Sample Outputs

Key Metrics

• Total Warehouses: 8
• Total SKUs: 32
• Shortage %: 25.0%
• Average SPI: -5.2
• Potential Cost Saving: ₹15,250

Transfer Recommendations

Product_Category	From_Location	To_Location	Units	Estimated_Saving_INR
Electronics	Bangalore	Delhi	30	240.00
Clothing	Mumbai	Pune	25	125.00
Books	Chennai	Kolkata	15	30.00

Screenshots

Main Dashboard

Stock Pressure Index

Transfer Network

Troubleshooting

Common Issues

1. File Upload Errors

   • Ensure CSV files have correct column names
   • Check date format (YYYY-MM-DD)
   • Verify numeric columns don't contain text

2. No Recommendations Generated

   • Check if warehouses have both surplus and shortage
   • Verify product categories match between files
   • Ensure sufficient data for demand calculation

3. Performance Issues

   • Large datasets may take time to process
   • Consider filtering data for testing
   • Use sample data for initial testing

Data Validation

The application automatically validates:

• Required columns presence
• Negative stock or cost values
• Date format consistency
• Data type compatibility

Advanced Features

Linear Optimization (Future Enhancement)

The system is designed to support PuLP-based linear optimization:

• Minimize total transfer costs
• Subject to stock constraints
• Consider distance and transportation costs

Custom Distance Matrix

For production use, implement:

• Real distance calculations between locations
• Transportation cost factors
• Delivery time constraints

Support

For issues or questions:

1. Check the troubleshooting section
2. Verify your data format matches requirements
3. Test with sample data first
4. Check console for error messages

Updates

Version 1.0

• Initial release with core functionality
• Sample data and basic optimization
• Interactive dashboards and what-if analysis

Planned Features

• Advanced optimization algorithms
• Real-time data integration
• Multi-objective optimization
• Machine learning demand forecasting

---

Built with Streamlit, Pandas, and Plotly