2016 INFORMS Annual Meeting Program

SB05

INFORMS Nashville – 2016

SB05 101E-MCC Real Options in the Energy Sector Sponsored: Energy, Natural Res & the Environment, Energy I Electricity Sponsored Session

SB06 102A-MCC Panel: What Industry Wants Analytics Graduates to Know Sponsored: Data Mining Sponsored Session Moderator: Thomas Tiahrt, The University of South Dakota, Beacom School of Business, Rm 229, Vermillion, SD, 57069, United States, Thomas.Tiahrt@usd.edu A panel discussion on What Industry Wants Analytics Graduates to Know 1 - What Industry Wants Analytics Graduates To Know Panelist: Eric B Stephens, Vanderbilt University Medical Center, eric.b.stephens@vanderbilt.edu 2 - What Industry Wants Analytics Graduates To Know Panelist: Sean T MacDermant, International Paper, yorrie@macdart.com 3 - What Industry Wants Analytics Graduates To Know Panelist: Alkis Vazacopoulos, Optimization Direct, Inc., alkis@optimizationdirect.com SB07 102B-MCC Data Mining in Decision Analytics: Predictive Modeling in Theory and Applications Sponsored: Data Mining Sponsored Session Chair: Ali Dag, Auburn University, Auburn, AL, United States, azd0033@auburn.edu 1 - A Novel Sentiment Analytic Methodology For Multinomial The objective of this study is to classify the customer reviews (on a five-star scale) that were collected for 3 different product/service. To achieve this goal, a novel classification framework is built by constructing a unique predictor, which includes rich information gathered by using all of the extracted features. The results indicate that the proposed method outperforms the alternatives. 2 - Probabilistic Decision Analytic Risk Level Prediction Model For Kidney Transplants Kazim Topuz, Wichita State University, Wichita, KS, United States, mktopuz@gmail.com, Mehmet B Yildirim, Ferhat Zengul, Ali Dag The objective of this study is to define risk levels and offer additional insights into the factors affecting the short, medium and long-term success/failure of a kidney transplant from deceased donor by using machine learning techniques. We utilized an exhaustive variable selection algorithm to eliminate improper/noisy variables by combining medical knowledge and mathematical models on large pool of variables. Then we employed BBN to extract the hidden patterns and relations between predictor variables as well as multi-class response variable. 3 - Ensemble Model With Cluster Analysis For Short-term Stock Prediction Bin Weng, Auburn University, Auburn, AL, 36849, United States, bzw0018@auburn.edu, Fadel Mounir Megahed, Chen Li The stock market is one of the most important ways for companies and individuals to raise money due to the feature of publicity and high liquidity. Accurately predicting stock market is extremely difficult due to the non-linear, volatile and complex of the market. The purpose of this study is to develop a model to predict stock’s short-term returns using disparate data sources from online data, economic data, technical indicators, and traditional history data. This study uses cluster analysis to cluster the trading days into different time periods and ensemble machine learning methods to develop the models for each period. As a result, the overall prediction performance has been increased. Classification Of Product And Service Reviews Ali Dag, Auburn University, azd0033@auburn.edu

Chair: Stein-Erik Fleten, Norwegian University of Science & Technology, NO-7491 Trondheim, Trondheim, NO-7491, Norway, stein-erik.fleten@iot.ntnu.no 1 - Switching From Oil To Gas Production – Decisions In A Field’s Tail Production Phase Kristian Støre, Norwegian University of Science and Technology, Bodø, Norway, krm@uin.no, Verena Hagspiel, Stein-Erik Fleten, Claudia Nunes We derive an optimal decision rule with regards to making an irreversible switch from oil to gas production. Assuming that both the oil and gas prices follow a geometric Brownian motion we derive a quasi-analytical solution for the exercise threshold. We also derive the related abandonment option. When comparing the use of a static decision rule to the proposed option approach we show that the value loss can be substantial for the abandonment option. For the switching option we find that with low gas prices the value loss can be more substantial than for the abandonment option, while for high gas prices it may be optimal to switch even as oil production is still generating positive cash flows. 2 - Resilience And Investment Valuation Of A Microgrid: A Real Options Approach Reinhard Madlener, Full Professor of Energy Economics and Management, RWTH Aachen University, E.ON Energy Research In this study, microgrids are discussed as a possible decentralized system approach to stabilize local power supply. Microgrids are a way to achieve a higher resilience for a whole energy supply system. We introduce and empirically apply a definition and quantification method for the resilience of a microgrid. Investment feasibility of the installation of different combinations of components is evaluated by adopting a real options approach for the optimal time to invest that takes the uncertainty about future developments into account. 3 - Real Options In Renewable Portfolio Standards Ryuta Takashima, Tokyo University of Science, takashima@rs.tus.ac.jp, Makoto Goto In order to promote renewable energy generation, the schemes as renewable portfolio standards have been introduced in various countries. Thus the power generators make investment decisions allowing not only for uncertain demands and competitors’ strategies but also for the schemes. In this work we model an equilibrium investment strategy of generators to analyze an effect of the schemes on the investment in competitive electricity market. The market is composed of non-renewable and renewable sectors. We show how the uncertainty affects the investment timing for both generators with the scheme. 4 - Structural Empirical Analysis Of Hydropower Scheduling Stein-Erik Fleten, Norwegian University of Science & Technology, stein-erik.fleten@iot.ntnu.no, Maren Boger, Jussi Keppo, Alois Pichler, Einar M Vestbøstad Our goal is to study how price expectations are formed in an electricity market. In the context of a single hydropower producer in the Nordic market, we expect the forward curve to have a strong influence. The alternative we allow for is a seasonal autoregressive joint inflow and spot price model that takes dry- and wet year dynamics into account. Using observed time series of generation, reservoir trajectories and technical plant data, and a structural model of optimal releases, our initial findings indicate that forward prices have influence on price expectations. An important byproduct of the proposed procedure is estimates of marginal water values. Center, FCN, Mathieustrasse 10, Aachen, Germany, RMadlener@eonerc.rwth-aachen.de, Lisa Goebbels

44