ketannnn/Retail-Sales-Intelligence-Dashboard

🛒 Retail Sales Intelligence Dashboard

A complete Data Analyst portfolio project — data cleaning, SQL analysis, EDA, and an interactive Streamlit dashboard built on the Superstore sales dataset.

📌 Problem Statement

A Superstore Giant is experiencing intense market competition and seeks data-driven guidance on:

• —Which products, regions, categories and customer segments to prioritise or avoid
• —How discount policies affect profitability
• —Which shipping modes are most cost-effective
• —Where consistent losses are occurring and why

📂 Dataset Description

Source: Sample – Superstore public dataset (9,994 rows, 21 columns)

🛠️ Tools Used

🗂️ Project Structure

retail-sales-intelligence-dashboard/
├── data/
│   ├── raw/
│   │   └── Sample - Superstore.csv
│   └── processed/              ← auto-generated at container start
│       └── cleaned_superstore.csv
├── .streamlit/
│   └── config.toml             ← Streamlit server config for HF Spaces
├── notebooks/
│   └── 01_data_cleaning_and_eda.ipynb
├── sql/
│   └── business_queries.sql
├── src/
│   ├── data_cleaning.py
│   ├── database.py
│   └── analysis.py
├── dashboard/
│   └── app.py
├── Dockerfile                  ← HF Spaces Docker deployment
├── requirements.txt
├── README.md
└── .gitignore

✨ Key Features

• —Automated Data Cleaning — snake_case normalisation, datetime parsing, duplicate removal, safe null handling
• —Feature Engineering — order_year, order_month, order_quarter, delivery_days, profit_margin, discount_range, is_profitable
• —SQLite Integration — cleaned data stored in retail_sales.db for SQL-based analysis
• —10 SQL Business Queries — KPIs, trends, region/category/segment/product analysis, discount impact, loss detection
• —Interactive Dashboard — 8 dashboard sections with 15+ Plotly charts, KPI cards, sidebar filters, and CSV export
• —Dynamic Business Recommendations — auto-generated insights from the filtered data
• —Docker Deployment — runs on Hugging Face Spaces via Docker SDK

📊 Dashboard Sections

🔍 SQL Analysis Included

1. 1.Overall KPIs
2. 2.Monthly sales & profit trend
3. 3.Region-wise performance
4. 4.Category-wise performance
5. 5.Top 10 products by profit
6. 6.Bottom 10 products by profit (loss makers)
7. 7.Discount range impact on profit
8. 8.Ship mode performance
9. 9.Segment-wise performance
10. 10.Loss-making sub-categories
11. 11.Bonus: High-discount loss-making orders

💡 Key Business Insights

1. 1.Technology generates the highest profit margin; Furniture is the weakest.
2. 2.Tables and Bookcases are consistent loss-makers, especially at high discounts.
3. 3.Discounts above 20% almost always result in losses across all categories.
4. 4.The West region is the top revenue driver; the Central region needs margin improvement.
5. 5.Standard Class accounts for ~60% of orders but has the longest delivery times.
6. 6.Corporate and Consumer segments dominate sales volume; Home Office is underserved.

🚀 How to Run Locally

# 1. Clone the repo
git clone https://github.com/yourusername/retail-sales-intelligence-dashboard.git
cd retail-sales-intelligence-dashboard

# 2. Create and activate virtual environment
python -m venv venv
# Windows:
venv\Scripts\activate
# macOS/Linux:
source venv/bin/activate

# 3. Install dependencies
pip install -r requirements.txt

# 4. Clean the raw data
python src/data_cleaning.py

# 5. Load into SQLite
python src/database.py

# 6. Launch the dashboard
streamlit run dashboard/app.py

# 7. (Optional) Open the Jupyter notebook
jupyter notebook notebooks/01_data_cleaning_and_eda.ipynb

🐳 Run with Docker

# Build the image
docker build -t retail-sales-dashboard .

# Run the container
docker run -p 8501:8501 retail-sales-dashboard

Then open http://localhost:8501

📋 Resume Bullet Points

• —Built an end-to-end retail sales analytics project in Python using Pandas, SQLite, and Streamlit, analysing 9,994 transactions across 4 regions and 17 product sub-categories
• —Engineered 7 derived features (profit margin, delivery days, discount buckets) and automated data cleaning with encoding-safe CSV ingestion
• —Authored 10 SQL business queries identifying recoverable losses from high-discount unprofitable orders
• —Developed an interactive Streamlit dashboard with 15+ Plotly charts, dynamic KPI cards, sidebar filters, and a CSV export feature
• —Containerised the application with Docker and deployed to Hugging Face Spaces for public access

📄 License

MIT — free to use for portfolio and educational purposes.