Creative Commons

This dissertation was written as a part of the MSc in e-Business & Digital Marketing at the International Hellenic University. With the advent of the Internet a significant change was occurred in the way individuals purchase goods and services. More and more customers tend to read products reviews before making a purchasing decision. Thus, such reviews have become an important component of e-commerce websites. Moreover, online consumers’ expectations are continuous being increased over the years. These heightened customers’ expectations have increased the complexity of online systems that businesses need to operate. A dynamic approach for companies is needed in order to remain competitive and achieve customer satisfaction. Thus, big challenges of today’s businesses are how to successfully handle the huge amount of data produced by customers daily as well as how to mine valuable information from them to understand users’ preferences and make accurate recommendations. This thesis explores how companies can manage and extract useful information from the data available in their warehouses in e-commerce environment. More specifically, predictive analytics and machine learning algorithms are implemented, creating three different models that predict customer review rating score after a product purchase. Implementing data mining tools, firms can predict future customer behaviors and trends. In this way, e-retailers have the ability to make proactive, knowledge-driven decisions. In the first part of the current dissertation, the literature review is presented. On the second part, deep analysis of a dataset will be conducted in order to discover useful patterns and insights. The dataset used comes from the Olist Shops which is a platform that connects e-retailers with potential customers in Brazil. Finally, the predictive models will be built to determine the most important factors affecting the rating scores provided by reviewers.