2016 INFORMS Annual Meeting Program

171 / 561

INFORMS Nashville – 2016
171
4 - A Preference-Based Evolutionary Algorithm For Bi-Objective UAV
Route Planning In Continuous Space
Erdi Da demir, Hacettepe University, Ankara, Turkey,
, Murat Köksalan,
Diclehan Tezcaner Öztürk
We address the bi-objective route planning problem for unmanned air vehicles
that move in a continuous terrain. We develop a preference-based evolutionary
algorithm that converge the desired regions of the Pareto front using the route
planner’s preferences. The algorithm determines both the visiting order of the
targets and the specific trajectory between targets. Experiments show that the
algorithm works well.
MB65
Mockingbird 1- Omni
Field Experiment Research on Mobile and Social
Applications
Sponsored: Information Systems
Sponsored Session
Chair: Tianshu Sun, University of Southern California, 3670 Trousdale
Pkwy, Los Angeles, CA, 90089, United States,
1 - Evaluating Consumer m-Health Services For User Engagement
And Health Promotion
Vibhanshu Abhishek, Carnegie Mellon University,
, Rema Padman, Yi-Chin Lin
Mobile apps have great potential to deliver promising health-related interventions
to engage consumers and change their behaviors such as healthy eating. This
study proposes and evaluates three mobile-enabled interventions to address these
challenges: a mobile-based visual diary, image-based dietitian support, and peer
engagement. We examined their effects on user engagement and food choices via
a 4-month randomized field experiment and show a positive impact of the mobile
diary and dietitian support on improving customer engagement. Specifically, the
mobile-based visual diary and dietitian support each increases the log-odds ratio
of user engagement by 43.8% and 50.7%, respectively.
2 - The Effect Of Product Placement On Shopping Behavior At The
Point Of Purchase: Evidence From A Randomized Experiment
Using Video Tracking In A Physical Bookstore
Pedro Ferreira, Carnegie Mellon University,
Qiwei Han, João Paulo Costeira
Physical retailers are increasingly trying to understand in-store shopping behav-
ior in order to increase sales. However, measuring and analyzing shopper behav-
ior at the point of purchase in physical retailing remains challenging. In this
paper, we implement an in-vivo randomized field experiment in a physical book-
store. We leverage video tracking technologies to monitor how shoppers respond
to random book placement, which induces random search costs. More specifical-
ly, we randomize the position of newly released books on the top of a large table
with several rows and columns such that each book’s search cost becomes inde-
pendent of the book’s characteristics. We use advanced 3D cameras and vision-
understanding algorithms that can track human motions in real-time to over-
come the large costs associated to large-scale video data. This way we are able to
significantly reduce the cost of encoding shopper activities by more than 80%.
Our experimental results show that on an average day books placed at the edge
of the table are both picked and taken more often by consumers than books
placed in the center of the table. However, the likelihood of taking a book that
was picked is on average similar for books placed at the edge and at the center of
the table, that is, books at the edge of the table sell more only because they are,
on average, picked more often. Armed with this knowledge, the bookstore man-
ager may maximize profit by placing books with higher margins at the edge of
the table.
3 - Stimulating UGC Contribution Via Performance Feedback:
A Randomized Mobile Field Experiment
Yili Hong, Arizona State University, 832 W Wagner Dr, Gilbert, AZ,
85233, United States, Bin Gu, Chen Liang,
Gordon Burtch, Nina Huang
This study analyzes the effect of performance feedback on user content generation
through a randomized mobile experiment. We find heterogeneous treatment
effects that depend upon a subject’s gender and the framing of the feedback
supplied (altruistic vs. competitive). Specifically, we found that female users are
more responsive to altruism-framed performance feedback, whereas male users
are more responsive to competitively-framed performance feedback.
4 - Motivating Mobile App Adoption: Evidence From A Large-scale
Field Experiment
Tianshu Sun, University of Southern California, Los Angeles, CA,
United States, , Lanfei Shi,
Siva Viswanathan, Elena Zheleva
Using a randomized field experiment involving 250,000 customers, we investigate
1) whether and how a platform can motivate customers’ app adoption and 2) the
causal effect of induced mobile app adoptions on customer engagement. We find
that 1) providing information or incentives can both significantly increase
customers’ app adoption; 2) the effect of app adoption varies greatly depending
on how customers are motivated. Providing incentives increases adoption but not
engagement. In contrast, providing information leads to effective mobile app
adoptions that sustainably increase customers’ engagement. We further look into
multi-channel browsing data to understand the effect of app adoption.
MB66
Mockingbird 2- Omni
Data Visualization
Sponsored: Quality, Statistics and Reliability
Sponsored Session
Chair: Fadel Mounir Megahed, Auburn University, 3301 L Shelby
Center Auburn University, Auburn, AL, 36849, United States,
1 - Stock Market Exploration And Prediction Through Visualization
Xing Wang, Auburn University, Auburn, AL, 36849, United States,
Bin Weng, Fadel Mounir Megahed
Stock Market prediction has attracted much attention from both business and
academia. To explore the insight from the stock market history data could help
investors to make their decision more efficiently. The purpose of this study is to
develop a tool for visualizing and predicting the stock market through data
mining methods. In this study, the tool is developed using Shiny R, which gives
users useful information via visualizing related data from disparate data sources to
assist investors to make decisions. By introducing our application, we focus on
two things, one is the data visualization system design, and the other is reactive
programming, both are emphasized in our development process.
2 - What We Learned From Visualizing 25 Years Of
Statistics Research?
Fadel Megahed, Miami University,
Theyab Alhwiti, Mohammad Alamdar Yazdi, Maria Weese,
L. Allison Jones-Farmer
The size and scope of the literature on statistics can be overwhelming, which
makes it difficult to identify emerging trends and see the relationships between
different developments. Visualization techniques, coupled with statistical and data
mining methods, have been found effective in achieving these goals in a number
of application domains including healthcare and manufacturing research. In this
paper, we apply these concepts to the field of statistical sciences. Our dataset is
based on bibliographic information, including: authors, keywords, abstracts,
citations, and funding information, extracted from 10,030 papers published in the
17 ASA journals in the period of 1991-2015.
MB67
Mockingbird 3- Omni
Advances in Degradation Modeling and Operations
Management
Sponsored: Quality, Statistics and Reliability
Sponsored Session
Chair: Xiao Liu, IBM T.J. Watson Research Center, 1101 Kitchwan
Road, Yorktown Heights, NY, 10598, United States,
1 - Kalman Filter Based Logistic Regression For
Degradation Analysis
Erotokritos Skordilis, University of Miami, Miami, FL,
United States, , Ramin Moghaddass
We propose a new stochastic approach for degradation analysis using a
combination of Bayesian Filtering and Binary Classification that can transform
real-time condition monitoring signals to actionable insights. Analytical results for
important reliability measures (e.g. RUL) will be given and a closed-form solution
for the marginal log-likelihood function will be developed. Finally, a dynamic
cost-effective predictive maintenance policy based on the proposed degradation
structure will be introduced and its benefit over time-based preventive
maintenance and corrective maintenance policies will be presented with a set of
numerical experiments.
MB67