Identifying Frequently Cancelled Products and Understanding Why

This project analyses the Amazon Sales dataset to identify patterns in product cancellations. The objective is to determine which products are most frequently cancelled and why, providing actionable insights to reduce cancellation rates and optimise inventory management.

Problem Statement:

Frequent cancellations can impact revenue, disrupt inventory management, and negatively affect customer satisfaction. This project analyses Amazon Sales data to identify which products are most frequently cancelled and the underlying reasons for these cancellations.

Tools Used:

• Excel (Data Cleaning, Analysis, Correlation Analysis, Chi-square Test, Dashboard Building)
• Power BI (Data Visualisation)

Press enter or click to view image in full size

Steps Taken:

Data Cleaning:

• We began by cleaning the data, ensuring there were no duplicates.
• Removed unwanted columns that were irrelevant to the analysis, such as product description or customer address.
• Standardised field names to maintain consistency, e.g., changing "OrderID" to "Order ID" for clarity.
• Corrected data formats to ensure date columns were formatted as dates and numeric columns as numbers.
• Addressed inconsistent data by standardising values like "Cancelled" and "CANCELLED."
• Trimmed leading and trailing spaces to avoid data entry errors.

Press enter or click to view image in full size

Data Analysis:

• Using Excel, we conducted a correlation analysis to examine relationships between product categories and cancellation rates. The analysis yielded a value of 0.99, indicating a strong relationship.

• We also performed a Chi-square test in Excel to explore the independence between product categories and cancellation rates, which further validated the relationship.

Press enter or click to view image in full size

• We created Pivot Tables to aggregate cancellation counts by product category and order status, helping us identify key trends.
• We analyzed monthly cancellation trends to identify seasonal patterns.
• Using these insights, we identified the top cancelled products by volume and percentage, which helped prioritize further analysis.

Dashboard Creation:

Press enter or click to view image in full size

• We developed an interactive Excel dashboard that visualises cancellation rates across product categories and order statuses.
• Slicers were added to the dashboard for dynamic filtering by product category, month, and order status.
• We used visualisations like bar charts, line graphs, and data tables to present the findings clearly.
• The final dashboard was uploaded for stakeholder review and data-driven decision-making.

Key Findings:

• The correlation analysis and Chi-square test confirmed a strong relationship between certain product categories and higher cancellation rates.
• The kurta product category showed a notable spike in cancellation rates, prompting further investigation.
• We identified seasonal trends and shipping delays as potential causes of high cancellation rates in specific product categories.
• The dashboard allowed stakeholders to easily identify problematic products and trends in cancellations.

Recommendations:

• We recommend implementing targeted quality control measures for products with high cancellation rates, especially kurtas.
• Improving communication regarding shipping timelines will help manage customer expectations and reduce cancellations.
• Optimizing return policies can help minimize unnecessary cancellations while maintaining customer satisfaction.

Personal Note:

I started my data analytics journey in February 2025, following the encouragement of a senior colleague at work. Initially, I was sceptical about learning data analytics but here I am today, sharing my first project. It has been a combination of excitement, challenges, and a whole lot of learning. But through all of this, one thing has been clear: learning data analytics is a rewarding journey.

This project was done in collaboration with my partner, Omotayo Olokede — Thank you so much.