Time Sensitive Operations

Time is an essential component of many operations. Any good or service provided by its producers must be delivered on time. Advancements in information technology and urban development through years has only exacerbated the demand and importance of timely operations. For certain good and services being timely is a new feature that was not required before whereas for others being timely is at the core of their value proposition. This dissertation examines three problems that have arisen in response to these advancements in information technology and urban development. In the first chapter, we study the dispatching problem faced by the dispatchers in Emergency Medical Services (EMS). Decision making in 911 dispatching of EMS plays a critical role in potentially saving patients’ lives during time-sensitive emergencies: Getting resources to a patient as quickly as possible is essential, and any delays can be life-threatening. Motivated by the challenges faced by EMS dispatchers we develop a novel data-driven dispatching framework by combining ideas from Machine Learning and Optimization to help guide dispatcher decisions. First, we propose a novel omniscient optimization model for ambulance dispatching that incorporates forward-looking

decisions. Second, we employ a Scenario Based Robust Optimization framework that utilizes ideas from Stochastic Programming and Robust Optimization, where the uncertainty sets and associated scenarios are built using our novel Closest Neighbours Clustering method. This method first uses past data to select a set of similar calls to the current emergency call to build a set of future scenarios. Then it clusters the set of scenarios using the scenario metric we develop. These clusters are later fed into our Scenario Based Robust Optimization model to produce a dispatch decision under future uncertainty. Experimental results show that our framework can reduce the percentage

of late responses by as much as 25% compared to current dispatch methods. In the second chapter, we study an on-demand platform’s delay information disclosure policy

when the platform serves two classes of users—consumers and providers—who seek matches to each other using the platform. We model the platform as seeking to maximize the average rate at which these users are successfully matched by choosing one of three information regimes; occupancy

information regime (disclosing the current system occupancy to both user classes), and two asymmetric information regimes (disclosing no information to one user class and occupancy information to the other). Arriving users are strategic and decide whether to join the system or not based on the delay information that the platform provides. In our base model, we consider users of each class as being either patient (will wait to be matched) or impatient (will join only if they expect to be matched immediately). We find that depending on the parameter setting, any of the three

information disclosure policies could emerge as optimal; however, the optimal policy has a complicated dependence on the parameters. We analytically establish sufficient conditions driving the platform’s decision. We also examine two limiting settings: (i) patience profile discrepancy between the two user classes; and (ii) market size imbalance between the two user classes. For these two limiting settings, we characterize the platform’s optimal information regime and show that all three disclosure policies

may again emerge in equilibrium. We then numerically examine the impact of the platform’s choice of information regime on users’ welfare, and find that the platform’s choice can also maximizes the welfare of both user classes, but only if this choice is to disclose occupancy information to both classes. We extend our base model to study how the platform’s information regime choice changes when user patience levels are more heterogeneous. In the third chapter, we study the problem of balancing efficiency and risk in multi-class screening systems. Security screening systems aim to identify malevolent people and illicit goods.

But screening operations may also result in long wait times at checkpoints. Selecting appropriate screening procedures thus involves a trade-off between efficiency and risk. This is complicated by the heterogeneity of screening jobs (which pose various threat levels) and the strategic behaviors of

human agents (who may renege prior to screening if they are malevolent and perceived risk levels are too high). We apply a speed-quality trade-off perspective to security operations and extend the speed-quality trade-off literature to a multi-class setting with heterogeneous and strategic agents. From a practical standpoint, our work supports tactical decision-making for dynamically selecting

screening procedures, and strategic decision-making for designing pre-screening profiling programs. We formulate this problem using continuous-time infinite-horizon Markov decision processes to optimize service rates in an M=M=1 queue with heterogeneous jobs, as a function of observed queue lengths and a threat level estimate for each job. We propose an extension to capture endogenous strategic behaviors of heterogeneous agents, given information asymmetries between agents and the screening operator. We show that the optimal policy exhibits a double threshold behavior: the shorter the queue length and/or the larger the risk, the stricter the screening. Numerical results show

that leveraging job-level risk information can reduce expected costs by up to 6-7%, as compared to single-class decision-making schemes. Moreover, anticipating agents’ strategic behaviors results in more intensive screening in an attempt to force malevolent agents to renege. This reneging

behavior can lead to counter-intuitive results: Slower screening mitigates expected risks and may also, surprisingly, reduce expected queue lengths.