This end-to-end data analytics project transforms the US Superstore retail dataset — spanning 9,994 order records across 2014–2017 — into a fully actionable business intelligence system. The project covers the complete analytics lifecycle: raw data ingestion, SQL-driven cleaning and exploratory analysis, Python-based EDA and visualization, and an interactive Streamlit dashboard. Covering 793 unique customers, 1,862 products, and 5,009 orders across 49 US states, the project surfaces insights into revenue trends, profit leakage, discount governance, regional performance, and customer lifetime value — delivering concrete recommendations at each stage.
- Clean and standardise raw retail data using SQL and Python pipelines
- Conduct comprehensive EDA to profile revenue, profit, and order behaviour
- Answer 30+ business questions on segments, regions, products, and customers
- Identify discount-driven profit leakage and loss-making sub-categories
- Build an interactive Streamlit dashboard with real-time KPI monitoring
- Deliver ranked recommendations grounded in quantitative findings
The project spans SQL, Python, and a Streamlit application — each layer purpose-built for a distinct stage of the analytics workflow:
- Data cleaning — missing values, type conversion, standardisation, outlier flagging
- SQL EDA — category, segment, region, product, and shipping analysis
- Business query writing — 30+ annotated queries with insight and recommendation
- Python EDA — KPI snapshot, correlation heatmap, trend analysis, top-performer ranking
- Interactive dashboard development with Streamlit and Plotly
- Business storytelling — translating query results into prioritised recommendations
📁 superstore/
└── 📂 Data-Analytics-Project/
├── 📂 Dataset/
│ ├── Raw_Data.csv
│ └── Cleaned_Data.csv
├── 📂 Python/
│ ├── Data_Cleaning.ipynb
│ ├── EDA.ipynb
│ ├── Data_Visualization.ipynb
│ ├── Dashboard.py
│ └── Requirements.txt
└── 📂 SQL/
├── Data_Cleaning.sql
├── EDA.sql
└── Business_Queries.sql
A visual walkthrough of the analysis outputs — from Streamlit KPI dashboards and Plotly interactive charts to SQL query results and Python EDA notebooks.
Across 9,994 order records spanning four years, the analysis uncovered a total revenue of $2.3M and a net profit of $286K — a 12.5% margin. The West region led profit performance while certain sub-categories (notably Tables and Bookcases in Furniture) were found to be persistently loss-making. Loss-generating orders carried an average discount rate significantly above the profitable cohort, validating a clear discount-governance problem. The Streamlit dashboard enables dynamic filtering by date range, category, region, and segment — exposing KPI shifts in real time and powering geographic treemaps across 49 states and 531 cities.
This project demonstrates a production-ready analytics workflow applied to a real-world retail dataset. By combining rigorous SQL data cleaning and business querying with Python-based EDA and an interactive Streamlit dashboard, it delivers a reusable template for retail business intelligence. The layered approach — raw CSV → cleaned dataset → SQL insights → Python visualisations → live dashboard — illustrates how structured analytical thinking transforms raw transactional data into prioritised, stakeholder-ready recommendations covering pricing strategy, discount governance, regional investment, and customer retention.