Doctoral Dissertations

Date of Award


Degree Type


Degree Name

Doctor of Philosophy


Business Analytics

Major Professor

Wenjun Zhou

Committee Members

Emre Demirkaya, Kelly Hewett, Qing Liu


With the advancement of information technologies and social media, a greater proportion of customer data is generated from sources such as word-of-mouth, purchasing records, and latent profiling. These new data sources give businesses more opportunities to learn from data during the decision-making process. They can, for example, understand their customers' reactions to their products and forecast their future spending. Meanwhile, questions arise about how businesses can effectively use data to generate reliable insights. Traditional analytical tools are limited in their ability to obtain, process, and discover knowledge from such large amounts of data. As a result, it is now critical for researchers and practitioners to create new data-handling tools.

Machine learning, the study of computer systems that can learn and develop on their own through experience and data, is one of the emerging tools. Machine learning has already been used in many sectors of business and management, including automation, financial management, and user behavior analysis. Although its benefits have been widely demonstrated such as streamlining time-intensive data entry documentation, assisting in accurate sales forecasts, as well as segmenting customers and precisely projecting their lifetime value, there is still room for more trustworthy procedures in additional applications to be developed.

Driven by the increasing needs of machine learning for business analytics, this dissertation is divided into three chapters that demonstrate the problems and solutions for learning customers from a business analytical standpoint. The first chapter creates a text mining tool for analyzing online reviews in order to automatically infer reviewers' brand experiences and assist businesses in determining which product features should be improved in the future. The second chapter proposes a model for dealing with asynchronous and irregular time series data. The goal is for businesses to learn the online word-of-month with varying degrees of granularity in order to take more timely actions. The third chapter proposes an interpretable machine learner to explain key performance indicators (for example, customer spending) from various segments. With the model's assistance, managers can later develop targeted strategies.

Available for download on Monday, May 15, 2028

Files over 3MB may be slow to open. For best results, right-click and select "save as..."