Leveraging association rule mining to accelerate sales

Thumbnail Image
Journal Title
Journal ISSN
Volume Title
School of Business | Master's thesis
Degree programme
Information and Service Management (ISM)
Companies operating in challenging business environments, characterized by the proliferation of disruptive technologies and intensifying competition, are obliged to re-evaluate their strategic approach. This has become the norm in the retail industry and traditional brick-and-mortar stores. Particularly local market players with scarce resources are looking into alternative solutions to delivering a unique customer experience with the intention to preserve their profitability. Customer experience has been an integral topic within academic research for decades, and has also substantiated its value in pragmatic contexts. Recent developments in this field have triggered the constitution of customer experience management functions, which aim to adopt a holistic approach to the customer experience. This enforces a quantitative perspective highlighting the role of customer transaction data. Association analysis is one of the most well-known methodology used to detect underlying patterns hidden in large transaction data sets. It uses machine learning techniques to firstly identify frequently purchased product combinations and secondly, to discover concealed associations among the products. The association rules derived and evaluated during the process can potentially reveal implicit, yet interesting customer insight, which may translate into actionable implications. The practical consequences in the framework of this study are referred to as sales increasing strategies, namely targeted marketing, cross-selling and space management. This thesis uses Python programming language in Anaconda’s Jupyter Notebook environment to perform association analysis on customer transaction data provided by the case company. The Apriori algorithm is applied to constitute the frequent itemsets and generate association rules between these itemsets. The interestingness and actionability of the rules will be evaluated based on various scoring measures computed for each rule. The outcomes of this study contribute to finding interesting customer insight and actionable recommendations for the case company to support their success in demanding market conditions. Furthermore, this research describes and discusses the relative success factors from the theoretical point of view and demonstrates the process of association rule mining when applied to customer transaction data.
Thesis advisor
Halme, Merja
association analysis, association rules, machine learning, transaction data, customer experience, customer experience management, customer insight, targeted marketing
Other note