BDAIB Session 1
Online Retail Data Analysis
UCI Online Retail Dataset
1
BDAIB Session 1
Online Retail Data Analysis
UCI Online Retail Dataset
1
1. Introduction & Objective
Analyse the UCI Online Retail dataset to identify useful business insights.
Focus areas: revenue, products, customers, countries, monthly trends and data quality.
Revenue = Quantity × Unit Price.
Validate unusual transactions before using them for business decisions.
Key principle: accurate analysis depends on clean, validated data.
2
2. Data Overview
UK-based non-store online retailer transaction dataset.
Analysis period: 1 December 2010 – 9 December 2011.
Key fields: Invoice number, Stock code, Product description, Quantity, Invoice date, Unit price, Customer ID and Country.
Checks included cancellations, missing customer information and unusual transactions.
3
3. Top Products by Revenue
Key finding: REGENCY CAKESTAND 3 TIER generated the highest recorded product revenue.
4
4. Top Customers by Revenue
Customer 14646 generated the highest recorded revenue. Missing Customer IDs limit completeness.
5
5. Revenue by Country
Key finding: The United Kingdom was the dominant revenue-generating market.
6
6. Monthly Revenue Trend
*December 2011 is incomplete because the dataset ends on 9 December. November was the highest month.
7
7. Data Quality & Validation
135,080 transactions have missing Customer IDs.
9,287 cancellation invoices were identified.
10,623 rows contain negative quantities.
The dataset ends on 9 December 2011, so December revenue is incomplete.
Large or unusual transactions require validation before being treated as genuine revenue.
8
8. Important Transaction Anomaly
Initial transaction revenue: £168,469.60.
Product: PAPER CRAFT , LITTLE BIRDIE.
Quantity: 80,995 | Unit price: £2.08.
Further investigation showed the transaction was associated with a cancellation.
Customer 16446 had only £2.90 in actual remaining purchases.
Lesson: automated outputs should be verified against the underlying transaction history.
9
Thank you