Amazon

Online shoppers often rely on ratings, reviews, and discounts to make decisions — but how reliable are these signals? In this project, I analyse an Amazon products dataset to explore:

1. How review volume affects rating reliability
2. Whether heavy discounts imply poor product quality
3. The relationship between price, discount, and customer satisfaction
4. Category-level pricing and rating behaviour

The project emphasises business-driven EDA, not prediction.

🎯 Objectives

1. Perform structured exploratory data analysis using Python
2. Identify counterintuitive patterns in customer behaviour
3. Visualise relationships between price, discount, rating, and review volume
4. Translate insights into business-relevant conclusions

🧾 Dataset Information

Source: Kaggle

Format: .xlsx

Each row represents a product review/listing

Key columns used:

product_name category actual_price discounted_price discount_percentage rating rating_count

🛠️ Tools & Libraries Used

Python Pandas (data manipulation & aggregation) Matplotlib & Seaborn (visualisation)

📊 Key Analyses & Visualizations

🔹 1. Rating Reliability vs Review Volume 🔹 2. Ratings of Highly Discounted Products (>50%) 🔹 3. Price, Rating & Discount Interaction 🔹 4. Category-Level Insights 🔹 5. Review Behaviour Analysis

🧠 Key Insights

1. ⭐ Ratings with low review counts are highly volatile

2. 📦 Customer trust increases with review volume

3. 💸 High discounts are often promotional, not quality-driven

4. 🏷️ Higher price does not always translate to higher ratings

📁 Project Structure

Amazon/ │ ├── data/ │ └── amazon_data.xlsx │ ├── notebooks/ │ └── Amazon.ipynb │ ├── visuals/ │ └── Amazon Visuals.ipynb │ ├── README.md

📊 EDA is crucial to assess data reliability, not just averages

📁 Project Structure