🛒 E-Commerce Intelligence Suite

The Smart Manager's Dashboard for Data-Driven Decisions

A powerful Streamlit-based analytics dashboard that helps e-commerce managers make smarter inventory and sales decisions through:

• 🛒 Cross-Selling Engine - Discover which products are bought together using Market Basket Analysis
• 📈 Inventory Forecaster - Predict future demand using ARIMA time series forecasting
• 📤 Upload Module - Analyze your own sales data with easy CSV upload

🚀 Quick Start

Prerequisites

• Python 3.8 or higher
• pip package manager
• Your own CSV sales data (see format requirements below)

🎯 Setup

1. Clone the Repository and naviagte to project directory:

git clone https://github.com/equaan/E-Commerce-Intelligence-Suite.git
cd E-Commerce-Intelligence-Suite

2. Create a virtual environment:

python -m venv venv
venv\Scripts\activate        # On Windows
source venv/bin/activate     # On macOS/Linux

3. Run the setup wizard:

python setup.py

🧩 What Happens Next

1. The setup script will:

   • ✅ Check Python version
   • ✅ Install all required dependencies
   • ✅ Configure your CSV file path
   • ✅ Initialize the database and start the app

2. ⚠️ Important: While installing dependencies, the process may pause for a few minutes.

   • Wait patiently — it’s downloading libraries in the background.
   • If it seems stuck for too long, open a new terminal and run:

     python setup.py

     again. The setup will continue from where it left off.

3. Once dependencies are installed, a data/ folder will be created.

   • Move your CSV file into that folder.
   • When prompted, enter the path as:

     data/your_file_name.csv

   • The setup will automatically detect the file, validate it, and begin initialization.

4. If initialization fails, the script will show manual recovery steps — just follow those instructions and re-run the command.

5. After successful initialization, the app will automatically launch in your browser.

📊 CSV Data Requirements

Your CSV file must contain these columns (exact names can be different, just update config.py):

Required Columns:

Column	Description	Example Values
InvoiceNo	Transaction/Order ID	"536365", "536366"
StockCode	Product ID/SKU	"85123A", "71053"
Description	Product name	"WHITE HANGING HEART T-LIGHT HOLDER"
Quantity	Number of items sold	6, 8, 2
InvoiceDate	Transaction date	"2010-12-01 08:26:00"
UnitPrice	Price per item	2.55, 3.39

Optional Columns:

Column	Description	Example Values
CustomerID	Customer identifier	"17850", "13047"
Country	Customer country	"United Kingdom", "France"

📁 Sample Datasets

Don't have your own data yet? Try these popular e-commerce datasets:

🛍️ Online Retail Dataset (Recommended)

• Source: UCI Machine Learning Repository
• Description: UK-based online retail transactions (2010-2011)
• Size: ~540K transactions
• Perfect for: Testing all features of the suite

🛒 E-commerce Data

• Source: Kaggle E-commerce Datasets
• Various datasets available with different formats
• Note: May require column mapping in config.py

🏪 Retail Sales Data

• Source: Kaggle Retail Analytics
• Multiple options for different retail scenarios
• Good for: Testing forecasting capabilities

📊 Custom Data Requirements

Your data should represent transactional sales records where each row is a product sold in a transaction. The system works best with:

• Multiple products per transaction (for cross-selling analysis)
• Time series data spanning several months (for forecasting)
• Consistent product identifiers (for accurate analysis)

📊 Features (See USER_MANUAL.md for a complete feature overview)

🏠 Home Dashboard

• Business overview with key metrics
• Daily sales trends
• Top-performing products
• Key insights and analytics

🛒 Cross-Selling Engine

• Market Basket Analysis using Apriori algorithm
• Product recommendation system
• Association rules with confidence, lift, and support metrics
• Interactive parameter tuning
• Visual charts and insights

📈 Inventory Forecaster

• ARIMA-based demand forecasting
• Configurable forecast periods (7-90 days)
• Confidence intervals and trend analysis
• Stock level recommendations
• Model accuracy metrics (MAPE)

📤 Upload Your Data

• CSV file validation and processing
• Sample data format download
• Real-time data cleaning and analysis
• Support for custom datasets

🗂️ Project Structure

DWM Project/
├── app.py                      # Main Streamlit application
├── initialize_data.py          # Database initialization script
├── requirements.txt            # Python dependencies
├── README.md                   # This file
├── setup.py                    # Installs deps, cleans data, inits DB, starts app
├── config.py                   # column mapping & validation
├── USER_MANUAL.md              # Feature usage guide
├── models/                     # ML models
│   ├── market_basket.py        # Market Basket Analysis
│   └── inventory_forecaster.py # ARIMA Forecasting
└── utils/                      # Utility modules
    ├── database_setup.py       # Database management
    └── data_processing.py      # Data cleaning & validation

🔧 Technical Details

Tech Stack

• Frontend: Streamlit, Plotly
• Backend: Python, Pandas, NumPy
• ML Libraries: mlxtend (Apriori), statsmodels (ARIMA)
• Database: SQLite with star schema

Database Schema

• FactSales: Main transaction table
• DimProduct: Product dimension
• DimCustomer: Customer dimension
• DimDate: Date dimension

Algorithms Used

• Apriori Algorithm: For market basket analysis and association rules
• ARIMA Model: For time series forecasting with automatic parameter selection

📈 Performance Notes

• Optimized for datasets up to 100K transactions
• Uses caching for improved performance
• Responsive design for various screen sizes
• Real-time analysis with progress indicators

⚙️ Configuration Guide (Only use when you have different column names and different type of dataset)

Step 1: Update CSV Path

Edit config.py and change the CSV_FILE_PATH:

# Examples of different data sources:
CSV_FILE_PATH = "data/my_sales_data.csv"                    # Local file
CSV_FILE_PATH = "C:/Users/YourName/Downloads/sales.csv"     # Full path
CSV_FILE_PATH = "https://example.com/sales_data.csv"       # URL (if accessible)

Step 2: Map Your Column Names

If your CSV has different column names, update the COLUMN_MAPPING:

COLUMN_MAPPING = {
    'InvoiceNo': 'order_id',        # Your CSV has 'order_id' instead of 'InvoiceNo'
    'StockCode': 'product_sku',     # Your CSV has 'product_sku' instead of 'StockCode'
    'Description': 'product_name',   # Your CSV has 'product_name' instead of 'Description'
    'Quantity': 'qty',              # Your CSV has 'qty' instead of 'Quantity'
    'InvoiceDate': 'purchase_date', # Your CSV has 'purchase_date' instead of 'InvoiceDate'
    'UnitPrice': 'price',           # Your CSV has 'price' instead of 'UnitPrice'
    'CustomerID': 'customer_id',    # Your CSV has 'customer_id' instead of 'CustomerID'
}

Step 3: Adjust Data Validation (Optional)

Modify validation rules in config.py if needed:

VALIDATION_RULES = {
    'min_quantity': 1,              # Change to 1 if you don't want to include returns
    'min_price': 0.01,              # Minimum price threshold
    'max_price': 5000,              # Maximum price threshold (adjust for your products)
    'remove_cancelled_orders': True, # Set to False if you want to keep cancelled orders
}

👤 Author

• Mohammad Equaan Kacchi

🤝 Contributors

• Hussain Anajwala