ecommerce-eda

This project performs an Exploratory Data Analysis (EDA) on the Online Retail Dataset to understand its structure, discover patterns, detect anomalies, and extract insights.

Comment

🧠 Exploratory Data Analysis (EDA) – E-Commerce Sales Analysis

📋 Description

This project performs an Exploratory Data Analysis (EDA) on the Online Retail Dataset to understand its structure, discover patterns, detect anomalies, and extract insights.

The analysis includes:

  • Data cleaning and preprocessing
  • Statistical summary and visualization
  • Key insights and observations

📂 Repository Structure

├── data/               
│   ├── raw/            
│   │   └── e-commerce_data.csv    # Original raw dataset
│   └── cleaned_data.csv           # Cleaned and processed dataset
│
├── notebooks/                      # Jupyter notebooks with analysis
│
├── results/                        # Output figures or reports
│
├── environment.yml                 # Conda environment file
│
└── README.md                       # Project documentation

⚙️ Installation & Setup

Option 1 — Using Conda

conda env create -f environment.yml
conda activate ecommerce-eda

🧾 Dataset Information

  • Dataset Name: Online Retail
  • Source: https://www.kaggle.com/datasets/carrie1/ecommerce-data
  • Format: CSV
  • Rows & Columns: 541909 rows and 8 columns
  • Description: This dataset contains all the transactions ocurring between 01/12/2010 and 09/12/2011 for a UK-based and registered non-store online retail.

📊 Analysis Overview

  • 1️⃣ Data Cleaning: Handling nulls, duplicates, and data types
  • 2️⃣ Exploratory Data Analysis: Distributions, histograms, RFM segmentation and descriptive stats
  • 3️⃣ Visualizations: Key plots and insights
  • 5️⃣ Conclusions: Summary of findings and implications

🖼️ Example Visualization

Total Revenue

🚀 How to Run

After activating the environment, launch Jupyter Notebook:

jupyter notebook notebooks/01_data_prep_cleaning.ipynb

Then open the notebook and run all cells in order.

Notebooks in /notebooks must be run in this order:

  • 01_data_prep_cleaning.ipynb
  • 02_eda.ipynb

This project was developed using Visual Studio Code with a Jupyter Notebook.

📚 Tools & Libraries

  • Python 3.11
  • Pandas
  • NumPy
  • Matplotlib
  • Seaborn
  • Jupyter Notebook

🧩 Key Findings

💵 Sales Overview

  • Sales have increased over time, with a significant rise in October 2011.
  • The period from September to December has the highest revenue, while January, February and April have the lowest.
  • Tuesday is the day with the lowest sales, whereas Wednesday has the highest.
  • The United Kingdom is the country with the most sales, followed by the Netherlands, Ireland and Germany.
  • The average sale on the lowest sales day, Tuesday, is $20 USD, while on the highest, Wednesday, it is $32 USD.
  • The 13th day of the month has the lowest average sales ($24 USD), whereas the 14th has the highest average ($38 USD).

📦 Product Performance

  • The Top 10 Best-Selling Products come from the following categories: Kitchen, Home Decor, Party Supplies and Bags. The best-selling product is a cake stand.
  • The Top 10 Most Cancelled Products belong to the categories: Crafts & DIY, Kitchen, Home, Toys, Party Supplies. The most cancelled product is a paper craft.
  • The following products appear in both the Top 10 Best Sellers and the Top 10 Most Cancelled: Regency Cakestand 3 Tier and White Hanging Heart Tlight Holder.
  • The best sales month for each of the Top 10 Best-Selling Products is:
    • Jumbo Bag Red Retrospot: November
    • Party Bunting: May
    • Rabbit Night Light: November
    • Regency Cakestand 3 Tier: December
    • White Hanging Heart Tlight Holder: January
    • Assorted Colour Bird Ornament: August
    • Chilli Lights: November
    • Jumbo Bag Pink Polkadot: October
    • Paper Chain Kit 50s Christmas: November
    • Spotty Bunting: August

🧍‍♂️Customer Behavior & Segmentation

  • From the RFM Segmentation:
    • Most of the customers fall into the following segments: At risk, New customers and Potential loyalist.
    • The least common segments are: About to sleep, Champions and Lost.
  • Revenue per customer is highly right-skewed, with a median of approximately $1,053 USD.
  • Although there are many new customers (according to the RFM Segmentation), the number of new customers per month has decreased over time.
  • The returning customer rate (customers with more than one transaction) is 98%.
  • The average time between orders is approximately 50 days, excluding same-day orders, which are the most common.

🛒 Purchase Patterns

  • The basket size tipically ranges from 1 to 12 items.
  • Most transactions have a basket value between $1 and $30 USD.
  • Tuesday and Wednesday are the days with the highest number of transactions.
  • The busiest time period is from 12:00 to 15:00.

⚙️ Operational Efficiency

  • Cancellation rate is low, with a maximum of 2.0%, decreasing to 1.3% in November 2011.
  • Most customers have a low cancellation rate, ranging from 0% to 20%.

👤 Author

🪪 License

This project is released under the MIT License.

Built With

Share this project:

Updates