Clustering-Based Analysis of Clickstream Data for Customer Journey Characterization

This thesis explores the use of clickstream data to characterize customer journeys through clustering-based analysis. The study focuses on a dataset from a single product line, while aiming to develop a generalizable framework applicable to other products and contexts. After preprocessing and normal...

Descripción completa

Detalles Bibliográficos
Autor: Pérez Singla, Daniel
Tipo de recurso: tesis de maestría
Fecha de publicación:2025
País:España
Institución:Universitat Politècnica de Catalunya (UPC)
Repositorio:UPCommons. Portal del coneixement obert de la UPC
Idioma:inglés
OAI Identifier:oai:upcommons.upc.edu:2117/452753
Acceso en línea:https://hdl.handle.net/2117/452753
Access Level:acceso abierto
Palabra clave:Cluster analysis
Big data
Consumer behavior
Clickstream data
Customer journeys
Clustering analysis
Anàlisi de conglomerats
Dades massives
Consumidors--Conducta
Àrees temàtiques de la UPC::Informàtica::Sistemes d'informació
Descripción
Sumario:This thesis explores the use of clickstream data to characterize customer journeys through clustering-based analysis. The study focuses on a dataset from a single product line, while aiming to develop a generalizable framework applicable to other products and contexts. After preprocessing and normalizing raw activity data, exploratory analysis and feature engineering were conducted to represent user sessions. Unsupervised learning methods, primarily K-Means clustering, were applied to group sessions and reveal behavioural patterns. The resulting clusters highlight distinct modes of engagement, from exploratory browsing to purchase-oriented interactions, and are compared against existing journey frameworks without enforcing strict alignment. Despite limitations in data granularity, the analysis demonstrates the potential of clustering to uncover customer behaviour dynamics. Future work should enrich datasets with multi-product interactions, campaign data, and user feedback, while extending the framework with supervised models for real-time journey stage prediction and personalized recommendations.