Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

TA29

2 - Understanding Gaze Events and Brain Activities in Online-learning through Electroencephalography and Eye-tracking Yuzhi Sun, Oregon State University, 1500 SW Jefferson St., Corvallis, OR, 97331, United States, David Nembhard We investigated the degree to which gaze events and brain activities are predictive of individual online learning performance through 2-by-2 experiment. Through the research, the gaze events were taped and documented via eye-tracking, the brain activities were monitored and recorded by the electroencephalogram (EEG), and the quiz scores measured the learning performance. ANOVA models and SEM models were considered to examine the relationships. Notably, the models indicate that there exists a significant association between design factors, gaze event durations, and learning outcomes. Levels of correlation between brain activities and gaze events were also found. 3 - Optimal Weighting for Exam Composition Sam Ganzfried, Ganzfried Research, 1504 Bay Road, Apt. 1706, Miami Beach, FL, 33139, United States, Farzana Beente Yusuf A problem faced by many instructors is that of designing exams that accurately assess the abilities of the students. Typically, generic question scores are used based on rough approximation of the question difficulty and length. We describe a novel framework where algorithms from machine learning are used to modify the weights to optimize the exam scores, using the overall final score as a proxy for a student’s true ability. We show that significant error reduction can be obtained by our approach over standard weighting schemes. We make several new observations regarding the properties of the “good’’ and “bad’’ exam questions that can have an impact on the design of improved future evaluation methods. n TA28 North Bldg 221A Resilient Rail and Failure Prediction Sponsored: Railway Applications Sponsored Session Chair: Nikola Besinovic, Delft University of Technology, Delft, 2600CN, Netherlands 1 - Train Rescheduling and Circulation Planning in Case of Complete Blockade for an Urban Rail Transit Line Yihui Wang, Beijing Jiaotong University, Beijing, China, Lingyun Meng, Tao Tang, Bin Ning With the rapid development of urban rail transit system, the disruptions occur more frequently due to the uncertain factors, such as signal failures. Once a disruption is occurred, the passengers’ safety and travel efficiency are seriously affected. Moreover, this impact of the disruption may spread to the adjacent lines and even to the whole network. Train rescheduling has a great effect on evacuating passengers quickly and relieving the mismatch of transport capacity and traffic volume. To maintain the transport capacity as much as possible under the disruption scenarios, this paper considers the train rescheduling and circulation planning in case of complete blockade for an urban rail line. 2 - Modelling Resilience of Rail Transport Networks Nikola Besinovic, Delft University of Technology, Stevinweg 1, Delft, 2600CN, Netherlands In current railway system disruptions are inevitable and more, their number is expected to further increase making the system even more vulnerable. We propose resilience performance indicators (RPI), to accurately estimate system performance during disruptions. We consider both topological and operational causes such as infrastructure and rolling stock failures. In addition, the effects of both single and multiple disruptions are investigated. For this purpose, historical traffic data from the Netherlands is used. As an outcome, we determine the most efficient RPI and a set of the most critical disruptions, which can be used in evaluating and optimizing resilience of future railway systems. 3 - Predicting Locomotive Failures: A Machine Learning Approach Clark Cheng, Sr. Director Operations Res & Chief Data Scientist, Norfolk Southern Corporation, 1200 Peachtree Street NE, Mail Stop 171, Atlanta, GA, 30309, United States, Mabby Amouie, Ilya Lavrik Locomotive reliability is a mission-critical issue for freight railroads. A locomotive failure will likely cause train delay, line-of-road congestion, service disruption, and ripple effect across the rail network. In this presentation, we’ll describe a machine-learning approach to predicting locomotive failures using sensor data ingested from locomotives.

4 - Modeling and Predicting Recurrent Rail Defects Faeze Ghofrani, University at Buffalo, SUNY, Buffalo, NY, United States, Qing He, Reza Mohammadi Provided six-year data from a US Class I railroad, this study develops a robust methodology for predicting the recurrence of rail defects at the same location by using the conditional frailty cox model. The results of this study are useful for railroads to develop effective strategies for track corrective maintenance to reduce the recurrence of rail defects. 5 - Positioning and Coordination of Resources for Reliable Emergency Response to Railroad Incidents Yanfeng Ouyang, U. of Illinois at Urbana-Champaign, 205 N. Mathews Ave, 1209 Newmark Lab, MC-250, Urbana, IL, 61801, United States, Siyang Xie This talk discusses strategic positioning and allocation of emergency response resources that may be impacted by rail incidents. Mathematical models and solution techniques are developed to systematically analyze the emergency response system in case: (i) joint dispatches are needed from multiple locations, and (ii) the emergency response system itself is vulnerable; e.g. due to blockage of railroad crossings after the incident. n TA29 North Bldg 221B Market Design Approaches for Transportation Systems Sponsored: TSL/Urban Transportation Sponsored Session Chair: Changhyun Kwon, University of South Florida, Tampa, FL, 33620, United States 1 - Spatio-temporal Pricing for Ridesharing Platforms Hongyao Ma, Harvard University, 33 Oxford Street, MD 242, Cambridge, MA, 02138, United States, Fei Fang, David C. Parkes A challenge in dynamic pricing in ridesharing platforms is to set prices that are appropriately smooth in space and time, so that drivers will choose to accept their dispatched trips, rather than drive to another area or wait for a better trip. We introduce the Spatio-Temporal Pricing (STP) mechanism, which is subgame- perfect incentive compatible for drivers, and also welfare-optimal, envy-free, individually rational and budget balanced from any history onward. We prove that there can be no dominant-strategy mechanism with the same economic properties, and show via simulation that STP achieves significantly higher social welfare than a myopic pricing mechanism, where drivers have high regret. 2 - Combinatorial Auction with Bidder-defined Items for Fractional Ownership of Autonomous Vehicles Mahdi Takalloo, Tampa, FL, 33613, United States, Aigerim Bogyrbayeva, Hadi Charkhgard, Changhyun Kwon In this study, a combinatorial auction with bidder defined items is proposed to design a market for vehicle fractional ownership under autonomous vehicles.Considering spatial information of bidder, we formulate the winner determination problem for the proposed combinatorial auction market under both discrete- and continuous-time settings.We show that the continuous-time model is superior, in terms of social welfare maximization, to the discrete-time model.We provide a clique-based reformulation of the continuous-time model, for which we develop an efficient algorithm. 3 - Acyclic Mechanism Design for Freight Consolidation Wentao Zhang, 3715 McClintock Ave, GER 240, Los Angeles, CA, 90089, United States, Nelson A. Uhan, Maged M. Dessouky, Alejandro Toriello Freight consolidation is a logistics practice that improves the cost-effectiveness and efficiency of transportation operations, and also reduces energy consumption and carbon footprint. A “fair” shipping cost sharing scheme is indispensable to help establish and sustain the cooperation of a group of suppliers in freight consolidation. We design a truthful acyclic mechanism to solve the cost-sharing problem in a freight consolidation system. We study the budget-balance of the mechanism both theoretically and numerically. We also study the economic efficiency of our mechanism numerically to investigate its impact on social welfare. 4 - American Options-Based Collaboration Mechanism for Multiple Less-Than-Truckload Carriers Under Demand Uncertainty Choungryeol Lee, Purdue University, West Lafayette, IN, United States, Srinivas Peeta We propose an American freight options-based collaboration mechanism for multiple less-than-truckload carriers that provides the flexibility to exercise the options at any time up to the expiry date so as to enhance operational efficiency. The proposed mechanism can address demand uncertainty by allowing for cost- efficient capacity allocation adjustments depending on future demand realization. Results of numerical experiments show that the proposed mechanism enables the collaborating carriers to enhance capacity utilization and profitability, while leveraging excess capacity in a more cost-efficient manner.

253