DATA-DRIVEN DECISION MAKING IN ONLINE AND BRICK-AND-MORTAR RETAILING

Gihan Samodha Edirisinghe

This dissertation focuses on the general theme of data-driven decision making in online and brick-and-mortar retail. Chapter One proposes a dynamic shelf allocation-relocation scheme for rearranging storewide product allocations over time to maximize impulse buying behavior. The proposed method rearranges items based on customer behavior with the current arrangement. The method applies insights from association rule mining to group highly affine and profitable product pairs, optimize the assignment of departments to store aisles, and determine the optimal within-aisle space allocations for the products of each department. This strategic rearrangement technique consistently outperforms random shelf-space rearrangement and, in many instances, exceeds the profit potential of a more traditional static shelf space arrangement. In Chapter Two, we conduct a data-driven analysis of the profitability and customer effects of three free shipping policies commonly adopted by online retailers at present. Using a Python-based web crawler, we obtained a unique dataset containing data from more than 80,000 products listed on Amazon.com. We then analyzed the data using non-linear mixed integer programs under different free shipping policies to compare their profitability and effects on customers. We find that membership free shipping policies are more profitable than the other policies considered. We also present insights on setting optimal membership fee levels and free shipping thresholds under different conditions. Chapter Three develops a methodology that creates structural equations using regression techniques based on known optimal results. The procedure is useful during situations such as initial decision-making on network-wide safety stock levels where quick, robust answers are sufficient. In Chapter Four, we present workable strategies for rural colleges and instructors operating on limited budgets to hold highly interactive question and answer (Q&A) sessions with accomplished guest speakers from industry. We introduce an approach for the selection of high-quality guest speakers, which utilizes the detailed résumés of tens of thousands of college alumni listed on LinkedIn. We also illustrate an innovative Q&A format that uses free cloud-based services such as Google Forms and Google Sheets. These successfully implemented strategies can address many challenges faced by rural schools in attracting and accommodating quality guest speakers for engaging exchanges with students.

DATA-DRIVEN DECISION MAKING IN ONLINE AND BRICK-AND-MORTAR RETAILING

Files and links (1)

Abstract

Metrics

Details