Informs Annual Meeting 2017

MB32

INFORMS Houston – 2017

MB31

3 - A Process Mining Framework for Analyzing Learning Clickstream Data Kexin Yin, University of Delaware, Newark, DE, United States, yinkexin@udel.edu, Harry Wang Learning analytics is an emerging field of research that aims to utilize a wide range of educational data to establish a deep understanding of the learning processes and learner behavior. In this paper, we propose a process mining framework for analyzing large-scale learning clickstream data collected from a major US university’s learning management system. We address a number of modeling and analysis challenges from a process mining perspective and propose new concepts for process-centric learning analytics. Based on the process mining results, we conduct an experiment to see how social influence affects learning processes and results. 350F Studying Social Issues with Social Media Invited: Social Media Analytics Invited Session Chair: Jessica Clark, NYU Stern School of Business, Brooklyn, NY, 11217, United States, jclark@stern.nyu.edu 1 - Diffusion of Ineffectual User Innovation Vidya Haran, University of Illinois Urbana-Champaign, 1206 South Sixth Street, 350 Wohlers Hall, Champaign, IL, 61822, United States, vharan@illinois.edu, Sung Won Kim We examine why ineffectual user innovations diffuse and persist in online communities. We investigate the case of ‘tweaking’, an ineffectual yet popular user innovation, on babycenter.com. Tweaking is the use of enhancing digital pictures to determine the results of a pregnancy test earlier than using the test as instructed. Though the tweaked results are not considered valid by the scientific community, the tweaking community continues to grow. We analyze online discussion to posit that psychological safety, anxiety, desirability bias, source credibility and social proof are the important factors that keep ineffectual user innovations from getting rejected. 2 - Gender Inequalities in Online Doctor Ratings Aishwarya Shukla, PhD, University of Maryland Smith School of Business, College Park, MD, United States, adshukla@rhsmith.umd.edu, Michelle Dugas, Gordon (Guodong) Gao, Ritu Agarwal Despite findings that female physicians are, on average, more patient-centered and effective than male physicians, evidence that patients are more satisfied with female physicians is equivocal. We sought to further clarify the relationship between physician gender and patient satisfaction with a unique dataset of online doctor reviews from India. To this end, we examined differences in the rates of positive and negative recommendations for male and female physicians, and explored moderating factors such as gender match. We also conducted text analysis of the reviews for further insights. Implications for physician care and consumer decision-making are discussed. 3 - Can You Pay My Bills? Crowdfunding as Supplemental Income Lauren Rhue, Wake Forest School of Business, 1834 Wake Forest Rd, 212 Farrell Hall, Building 60, NC, 27106, United States, rhuela@wfu.edu This study examines the role of crowdfunding in supplementing individuals’ income by using data from GoFundMe. Regional differences in opportunities, political leanings, and income inequality influence the crowdfunding goals and the campaigns’ success rate. Areas with fewer resources may experience more campaigns per capita and lower success because there is greater need for extra income yet there are fewer affluent potential contributors. Also, because personal social connections are critical in fundraising, we investigate the use of social media as well as the linguistic and image elements of the campaigns. MB30

351A INFORMS Korea Chapter Sponsored: INFORMS Special Sessions Sponsored Session Chair: Chang Won Lee, Hanyang University, School of Business, 17 Haengdang Dong, Seoul, 133-791, Korea, Republic of, leecw@hanyang.ac.kr 1 - Humanitarian Supply Chain Management: Research Opportunities and Implementation Issues This study is to explore research opportunities and Implementation issues in Humanitarian Supply Chain Management (HSCM). This study is to review some issues on social impact of HSCM consequences. Implication perspectives in SCM in non-profit sector are not reviewed well compared to those in SCM for-profit contexts so that this study provides the cases of challenges and research opportunities. 2 - Effects of Hospital Occupancy on Patient Length of Stay and Patient Outcome Yunsik Choi, Clemson University, 100 Sirrine Hall, Clemson, SC, 29634-1305, United States, yunsikc@g.clemson.edu, Lawrence Fredendall, Aleda Roth, Babur De Los Santos, Michael Makowsky We examine the effect of hospital occupancy on time from admission to procedure start (segment 1) and the time from procedure end to discharge (segment 2). We investigate the effect of hospital occupancy on the completion of time segment 1 and 2 and effects of the completion time on 30-days readmission. 3 - Designing Energy Supply Chains for Energy Food and Flood Kwon Gi Mun, Fairleigh Dickinson University, Fort Lee, NJ, United States, kgmun@fdu.edu, Yao Zhao, Raza Ali Rafique The interconnected issues of Energy, Food, and Flood are among the most formidable challenges faced by developing countries. The development of hydropower has the potential to address all these issues in the same time. We identify the unique features, economies, and trade-offs in the hydro system development. We also provide the end-to-end and dynamic perspectives. Chang Won Lee, Hanyang University, School of Business, 17 Haengdang Dong, Seoul, 133-791, Korea, Republic of, leecw@hanyang.ac.kr

MB32

351B Forecast Aggregation Invited: Forecasting and Prediction Invited Session

Chair: Yael S. Grushka-Cockayne, Darden School of Business, University of Virginia, 100 Darden Boulevard, Charlottesville, VA,

22903, United States, GrushkaY@darden.virginia.edu 1 - Automating Judgement and Decision-making: Theory and Evidence from Resume Screening

Bo Cowgill, Columbia University, 11 Fox Meadow Road, Scarsdale, NY, 10583, United States, bo.cowgill@gmail.com

I develop a formal model of the comparative advantages of human judgement and machines in decision-making. I subsequently test these predictions in a field experiment in applying machine learning for hiring workers for white-collar team-production jobs. The marginal candidate picked by the machine (but not by human screeners) is +17\% more likely to pass a face-to-face interview with incumbent workers and receive a job offer, b) +15\% more likely to accept job offers when extended by the employer and c) 0.2$\sigma$-0.4$\sigma$ more productive once hired as employees. 2 - Efficient Probability Forecasting using Ensembles Eva Regnier, Naval Postgraduate School, Monterey, CA, United States, eregnier@nps.edu We investigate the efficiency of probability forecasts derived from model ensembles—output from multiple models or multiple runs of the same model—- and recommend approaches to improving ensemble probability forecasts.

175