🛒 Retail Sales Analysis using Python

📌 Project Overview

This project explores the Global Superstore dataset to uncover actionable insights about sales, profit, customers, and regional performance.

The goal is to perform end-to-end data analysis using Python — from cleaning and feature engineering to exploratory data analysis (EDA), statistical testing, and key insights generation.

---

🎯 Objectives

• Understand sales and profit distribution across categories, regions, and segments.
• Identify key drivers of profitability and loss-making products.
• Perform time-series trend analysis and customer behavior study.
• Use statistical tests to validate observed patterns.
• Derive data-backed recommendations for strategic business decisions.

---

🧩 Dataset Information

Source: Global Superstore Dataset (Kaggle)
Size: ~51,000 rows
Key Columns:

• Order ID, Order Date, Ship Date, Ship Mode
• Customer ID, Segment, Region
• Category, Sub-Category
• Sales, Profit, Quantity, Discount

---

⚙️ Tools & Libraries

• Python 3.x
• Jupyter Notebook
• pandas, numpy
• matplotlib, seaborn, plotly
• scipy.stats (for statistical testing)
• datetime
• pycountry

---

🧱 Project Structure

Retail-Sales-Analysis/
│
├─ data/
│ 	├─ global_superstore.csv
│ 	├─ global_superstore_clean.csv
│ 	└─ global_superstore_capped.csv
│
├─ notebook/
│ 	└─ retail_sales_eda.ipynb
│
├─ visuals/
│ 	├─ univariate/
│   │ 	├─ numerical_plots/
│   │	│	├─ delivery_days_plot.png
│   │	│	├─ discount_plot.png
│   │	│	├─ profit_margin_plot.png
│   │	│	├─ profit_plot.png
│   │	│	├─ quantity_plot.png
│   │	│	├─ sales_plot.png
│   │	│	└─ shipping_cost_plot.png
│   │	│
│   │ 	└─ categorical_plots/
│   │		├─ category_countplot.png
│   │		├─ market_countplot.png
│   │		├─ region_countplot.png
│   │		├─ segment_countplot.png
│   │		├─ ship_mode_countplot.png
│   │		└─ sub_category_countplot.png
│   │	
│ 	├─ bivariate/
│   │	├─ category_vs_profit.png
│   │	├─ category_vs_sales.png
│   │	├─ correlation_heatmap.png
│   │	├─ discount_vs_profit.png
│   │	├─ region_vs_category.png
│   │	├─ region_vs_profit.png
│   │	├─ segment_vs_profit.png
│   │	├─ segment_vs_ship_mode.png
│   │	├─ segment_vs_sales.png
│   │	├─ region_vs_profit.png
│   │	└─ shipmode_vs_deliverydays.png
│ 	│
│ 	├─ multivariate/
│   │	├─ profitabilityByRegionCategory.png
│   │	├─ region-category-subcategory.png
│   │	├─ SalesProfitBySegmentCategory.png
│   │	├─ top10customers.png
│   │	└─ top10ProductsByLoss.png
│   │	
│ 	├─ trend_analysis/
│   │	├─ category-sales-trend.png
│   │	├─ monthlytrend.png
│   │	├─ orderstrend.png
│   │	├─ yearlytrend.png
│   │	└─ year-month-trend.png
│   │
│ 	└─ advance_analysis/
│   	├─ AOVyearly.png
│   	├─ cohort.png
│   	├─ pareto.png
│   	├─ profitability.png
│   	└─ unique_customers.png
│  
├─ LICENSE	
├─ requirements.txt
└─ README.md

---

📊 Project Workflow

1. Setup & Imports

Import all required libraries and define configurations.

2. Data Loading & Validation

Load dataset and perform initial exploration (shape, dtypes, missing values).

3. Data Cleaning & Feature Engineering

• Convert date columns and remove extra spaces
• Handle missing values & duplicates
• Engineer features like Delivery_Days, Month_Year, etc.
• Detect and cap outliers for Sales, Profit, and Shipping_Cost.

4. Exploratory Data Analysis (EDA)

• Univariate: Distribution of sales, profit, discounts, categories.
• Bivariate: Relationship between discount, profit, and segment.
• Multivariate: Combined effect of region, category, and segment.
• Trend Analysis: Monthly and yearly sales & profit patterns.
• Customer Cohort: Retention and AOV per year.
• RFM & Pareto Analysis: Identify high-value customers and top contributors.
• Geographical Insights: Regional sales visualization.

5. Statistical Testing

To validate key findings:

• Shapiro–Wilk Test: Check normality of sales & profit.
• Correlation Test: Numeric relationships.
• Kruskal-Wallis Test: Profit differences by region, category, segment.
• Chi-Square Test: Association between categorical variables.
• Mann–Whitney U Test: Binary group comparisons (e.g., high vs low discount).

6. Key Insights & Recommendations

Summarized actionable findings based on analysis and statistics.

7. Conclusion

Wrap-up of project outcomes and business takeaways.

---

🔍 Key Insights

• Highest sales observed in Central region, but profitability varies across sub-categories.
• Furniture often shows high sales but low profit margins due to high shipping and discounts.
• Technology drives most of the profit, especially under Corporate and Home Office segments.
• Sales peak in Q4 each year, suggesting strong seasonal patterns (likely due to festive or year-end promotions).
• Discounts beyond 20% drastically reduce profit margins.
• Top 20% customers contribute ~80% of total sales (Pareto principle).
• No significant profit difference across regions (statistical validation supports this).

---

💡 Recommendations

• Optimize discount strategy — cap discounts near profitable thresholds (~15–20%).
• Focus marketing on top-performing segments (Technology & Office Supplies).
• Reduce shipping costs for bulky items (especially Furniture) via logistics partnerships.
• Leverage Q4 trend — plan stock and campaigns ahead of the sales surge.
• Customer retention programs for high-value customers identified in RFM analysis.
• Periodic data review — continuous tracking of profitability drivers.

---

🧠 Conclusion

This project demonstrates a complete retail data analytics pipeline using Python.
It combines cleaning, feature engineering, visualization, statistical inference, and business recommendations — bridging the gap between data and decision-making.

---

▶️ How to Run

1. Clone the repository

   git clone https://github.com/ajx7/Retail-Sales-Analysis.git

2. Navigate to the project directory

   cd Retail-Sales-Analysis

3. Install dependencies

   pip install -r requirements.txt

4. Run the Jupyter Notebook

   jupyter notebook notebooks/retail_sales_eda.ipynb

---

📖 Dataset Source

Global Superstore Dataset (publicly available on Kaggle): Global Superstore Dataset – Kaggle

---

⚖️ License

This project is licensed under the MIT License — free to use with attribution.