The rapid digital transformation of the retail sector has shifted consumer expectations, requiring sophisticated analytical frameworks to interpret the intricacies of shopping behavior. This thesis investigates the implementation of a digital twin architecture to simulate, monitor, and predict consumer decision-making processes within a smart retail environment. By developing a specialized web-based simulation tool, the "Picklist", this research overcomes the limitations of traditional observational studies, enabling the collection of high-resolution behavioral data.The central methodological pillar of this work lies in the development of an interpretable predictive modeling framework, designed to map the relationships between digital interaction metrics—such as hovering duration, path navigation, and selection latency—and the ultimate purchase probability. Empirical analysis shows that the integration of digital simulation with predictive modeling and clustering enables a structured interpretation of behavioral patterns across sessions.This study not only validates the effectiveness of the digital twin paradigm for retail research but also establishes a scalable framework that can be leveraged to refine omnichannel marketing strategies. Ultimately, this work provides a bridge between theoretical consumer psychology and practical engineering applications, offering a foundation for future advancements in adaptive retail environments.
La rapida trasformazione digitale del settore retail ha profondamente alterato le aspet tative del consumatore, rendendo indispensabili framework analitici avanzati in grado di interpretare la complessità dei processi decisionali d’acquisto. Questa tesi approfondisce la progettazione e l’implementazione di un’architettura di digital twin finalizzata a sim ulare, monitorare e prevedere il comportamento del consumatore in un ecosistema di smart retail. Attraverso lo sviluppo di una piattaforma di simulazione web-based propri etaria, denominata "Picklist", la presente ricerca supera i limiti degli studi osservazionali tradizionali, permettendo l’acquisizione di dati comportamentali ad alta risoluzione.Il pilastro metodologico del lavoro risiede nello sviluppo di un framework interpretabile di modellazione predittiva, finalizzato a descrivere le relazioni tra le metriche di interazione digitale — quali il tempo di sosta (hover time), la navigazione nel catalogo e le latenze di selezione — e la probabilità finale di acquisto. Le analisi empiriche condotte mostrano come l’integrazione tra simulazione digitale, modellazione predittiva e clustering consenta una lettura strutturata dei pattern comportamentali osservati nelle sessioni.Tale ricerca non solo convalida l’efficacia del paradigma del digital twin nel contesto del retail, ma definisce altresì un framework scalabile, applicabile per l’ottimizzazione delle strategie di marketing omnichannel. Il lavoro colma il divario tra la psicologia del consumatore e le applicazioni ingegneristiche, gettando le basi per lo sviluppo futuro di ambienti di vendita adattivi e personalizzati.
SmartRetail predictive analytics: full-stack digital twin and behavioral predictive modeling
DAMIANI, LEONARDO
2025/2026
Abstract
The rapid digital transformation of the retail sector has shifted consumer expectations, requiring sophisticated analytical frameworks to interpret the intricacies of shopping behavior. This thesis investigates the implementation of a digital twin architecture to simulate, monitor, and predict consumer decision-making processes within a smart retail environment. By developing a specialized web-based simulation tool, the "Picklist", this research overcomes the limitations of traditional observational studies, enabling the collection of high-resolution behavioral data.The central methodological pillar of this work lies in the development of an interpretable predictive modeling framework, designed to map the relationships between digital interaction metrics—such as hovering duration, path navigation, and selection latency—and the ultimate purchase probability. Empirical analysis shows that the integration of digital simulation with predictive modeling and clustering enables a structured interpretation of behavioral patterns across sessions.This study not only validates the effectiveness of the digital twin paradigm for retail research but also establishes a scalable framework that can be leveraged to refine omnichannel marketing strategies. Ultimately, this work provides a bridge between theoretical consumer psychology and practical engineering applications, offering a foundation for future advancements in adaptive retail environments.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_06_Damiani.pdf
accessibile in internet per tutti
Descrizione: SmartRetail Predictive Analytics: Full-Stack Digital Twin and Behavioral Predictive Modeling
Dimensione
7.14 MB
Formato
Adobe PDF
|
7.14 MB | Adobe PDF | Visualizza/Apri |
|
2026_06_Damiani_executive summary.pdf
accessibile in internet per tutti
Descrizione: Executive Summary of the Thesis SmartRetail Predictive Analytics: Full-Stack Digital Twin and Behav ioral Predictive Modeling
Dimensione
1.05 MB
Formato
Adobe PDF
|
1.05 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/259960