Objective
The project focuses on H&M's dress category and examines differences in customer value, activity, and repeat purchasing to support more targeted marketing decisions.
Data preparation
The transaction table was joined with product and customer information before filtering dress-related records. The resulting analytical sample contains 3,247,122 transactions, 609,964 customers, and 10,286 products.
Methods
Descriptive statistics establish the category context. RFM features capture customer value, and K-means separates price-sensitive, dormant, potential, and high-value groups. Marketing recommendations map directly to segment behavior through differentiated benefits, cadence, and content.
Reflection
The segments are useful for operational prioritization, but boundaries depend on the observation window, feature scaling, and the selected number of clusters. Retention validation or incremental experiments would be needed to test whether the proposed actions improve repeat purchase.