Amazon Analysis
Excel | Python | SQL | Power BI
Interactive Power BI Dashboard
Why this project?
As my first end-to-end analytics project, the primary goal was not to solve a specific business problem but to build and understand a complete data analytics pipeline. The dataset was chosen because it contained realistic e-commerce transactions, sales metrics, and fulfilment information, providing an opportunity to explore the types of challenges commonly faced by analysts. The dataset, over 130,000 rows, provided enough of a challenge in regard to data cleaning and analysis. Click here to see dataset download.
Methodology
Dataset Aquisition: Dataset was aquired form Kaggle, a free open source data provider. This particular dataset was not a real-world dataset, but was designed to look and feel like one, simulating various formatting and data normalisation issues which are common place in real world datasets.
Data Cleaning in Python (VSCode): Used Pandas to import, clean, and format the raw, messy e-commerce datasets to ensure consistent dates, numeric values, and column structures.
Database & Modeling (PostgreSQL): Loaded the cleaned data into Postgres as a database host. Wrote basic query to explore the data and preview some buisness questions. See all python and SQL code on my GitHub here.
Visualization & Analysis (Power BI): Connected Power BI directly to the Postgres database and visualised key executive metrics on a well layed out summary page.
Key Insights
Buisness Insights
The business generated approximately £770k in revenue from over 120k orders during Q2 2022.
Average order value remained relatively low at around £6.22, suggesting a high-volume, low-value sales model.
Revenue remained broadly consistent throughout the quarter, with no significant prolonged declines in sales performance.
Product sales were concentrated among a small number of leading categories, with "Set" products contributing the largest share of revenue ~50%.
Sales performance varied significantly across Indian states, highlighting regional differences in customer demand.
The majority of orders were successfully fulfilled, with cancellations and delivery exceptions representing a relatively small proportion of total orders.
Project Takeaways
Building an end-to-end analytics pipeline requires different tools to perform different roles: Python for data preparation, PostgreSQL for data storage and querying, and Power BI for business reporting.
Effective dashboard design focuses on answering key business questions rather than displaying every available metric.
This project provided practical experience in transforming raw transactional data into a dashboard capable of supporting business decision-making.