Abstract
With the advance of computing power and explosion of data, over the past two decades, we have witnessed the rise of empirical operations management (OM) researches from scarcity to an emerging stream of work. To merge theory with practice, empirical OM requires fundamental changes in the way we develop methodologies and analysis frameworks. In this dissertation, we broaden this up-rising stream on three topics with particular interests in healthcare and transportation. First, we use a discrete choice ex periment with latent class multinomial logistic regression to quantify how non-urgent patients’ choices between emergency department (ED) and General Practitioner (GP) are influenced by ED/GP attributes, patients’ demographics, their perception of sever ity, and a new GP-referral financial incentive scheme. Next, we propose an integrated framework that integrates empirical analysis with analytical models to investigate a heterogenous population’s choices on car ownership, usage and transportation mode at the presence of ride-haling. Finally, we adopt a deep neural network, Long Short Term Memory network, to predict future travel time from past data sequence on an expressway with a scrutiny of input data’s quality on prediction performance of the model