Phd Canditate :

E-mail :

Supervisor :

Anastasia Griva

e-mail

Assoc. Professor K. Pramatari

PhD Thesis Abstract

Retailers have long recognized the importance of utilizing business analytics to support decisions. Taking into consideration both technological advantages and retailers’ business needs, we developed an approach with a view to infer a wealth of consumer behavioral insights. This is a clustering-based framework that could be utilized to divide shoppers’ visits in segments. Existing approaches examine either the items shoppers interact with in a single visit, or all shopper visits together. Thus, we introduce, as the core of the proposed approach, the concept of a customized unit of analysis. We seek to understand the shopper behind the visit via examining groups of X continuous in time visits. The utility and the applicability of this approach have been demonstrated by using real data in various retail contexts. Namely we utilized point-of-sales (POS) data from physical and web stores of different grocery retail chains and a fortune 500 do-it-yourself (DIY) retailer, data derived from RFID-enabled smart fitting rooms in a fashion retail chain, and other sensor data. The proposed approach is useful for both an academic and practical perspective. Apart from its applicability in different retail contexts, the results demonstrate significant business impact. The derived shopper insights can further be used to support several decisions in the retail domain e.g. via offering tailor-made marketing, that suits the specific needs and preferences of the different segments.