Modeling product stickiness in a grocery store

Oražem, Maruša (author), Šubelj, Lovro (mentor)

In the business world, keeping customers coming back and staying loyal is crucial for making money. We often measure this loyalty through customer satisfaction and how often people buy from a store or brand. This master’s thesis introduces the innovative index called Stickiness Index, tailored to capture and quantify repetitive purchases of items in the grocery stores. Our primary aim is to construct a robust Stickiness Index that effectively models items with high stickiness—those frequently repurchased after initial acquisition. The research unfolds in three key stages, including 3 different approaches for calculating Stickiness Index. Each stage is upgraded version of the previous, from simple event counting to incorporating item frequencies in the equation. In each stage we reveal the results and rationalize them. Additional fourth stage is presented, where we look at different item frequencies and how they affect the Stickiness Index. Analyzing the results, including price dynamics, item categorization, and seasonality, informs our understanding of consumer behavior. Moreover, this study demonstrates the real-world utility of the Stickiness Index by integrating it into machine learning models and strategic guidance for decision-making including the Stickiness Index.

• Language: English
• Keywords: grocery store, Stickiness Index, machine learning, analysis
• Work type: Master's thesis/paper
• Typology: 2.09 - Master's Thesis
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2023
• PID: 20.500.12556/RUL-151595
• COBISS.SI-ID: 169463811
• Publication date in RUL: 11.10.2023

Secondary language

• Language: Slovenian
• Title: Modeliranje lepljivosti izdelkov v trgovini z živili

Abstract

V poslovnem svetu je ohranjanje stalnih strank in njihova zvestoba ključnega pomena za ustvarjanje dobička. To zvestobo pogosto merimo prek zadovoljstva strank in kako pogosto ljudje nakupujejo v določeni trgovini. Magistrsko delo uvaja nov indeks, imenovan indeks lepljivosti (Stickiness Index), ki je prilagojen za ocenjevanje ponavljajočih se nakupov izdelkov v trgovinah s prehrambnimi izdelki. Naš osnovni cilj je izdelava indeksa lepljivosti, ki modelira izdelke z visoko lepljivostjo — tiste, ki se pogosto znova kupujejo po prvem nakupu. Raziskava se odvija v treh ključnih fazah, vključno z različnimi pristopi za izračun indeksa lepljivosti. Vsaka faza predstavlja nadgradnjo prejšnje, pri čemer začnemo s preprostim štetjem dogodkov in nadaljujemo z vključevanjem frekvence nakupovanja izdelkov v enačbo. V vsaki fazi predstavimo rezultate in jih pojasnimo. Poleg tega je predstavljena še dodatna četrta faza, kjer preučimo različne frekvence izdelkov in kako vplivajo na indeks lepljivosti. Analiza rezultatov, vključno s cenovno dinamiko, razvrščanjem izdelkov v skupine in sezonskimi vzorci, poglobi naše razumevanje potrošnikov. Poleg tega ta študija demonstrira praktično uporabnost indeksa lepljivosti z vključitvijo v modele strojnega učenja in usmerjanjem strateških odločitev v trgovini.

• Keywords: trgovina z živili, indeks lepljivosti, strojno učenje, analiza