2016 INFORMS Annual Meeting Program

MA54

INFORMS Nashville – 2016

MA54 Music Row 2- Omni Analytics and Operations Research for the IT Services Industry Sponsored: Service Science Sponsored Session Chair: Aly Megahed, IBM Research - Almaden, 650 Harry Road - Office D3-428, San Jose, CA, 95120, United States, aly.megahed@us.ibm.com 1 - Maximum Accuracy Is Not Always Optimal Ray Strong, IBM, San Jose, CA, United States, hrstrong@us.ibm.com, Aly Megahed, Janet Blomberg, Pablo Pesce, Yasuharu Katsuno, Sunhwan Lee The optimal features of a machine learning classifier or prediction model depend on the users of the analytics results. When the users have responsibility for acting on the results, as in the case of a sales force for cloud services, it is often more important to produce simple, understandable, and credible rules than to optimize for best prediction accuracy. 2 - An Optimization Approach To Revenue Forecasting In Multi-Staged Sales Pipelines Aly Megahed, IBM, San Jose, CA, United States, aly.megahed@us.ibm.com, Peifeng Yin, Hamid Reza Motahari Nezhad Services organizations manage a pipeline of sales opportunities with variable engagement lifespans and contract values. Accurate forecasting of contract signings by the end of a time period (e.g., a quarter) is vital for such organizations to effectively manage the pipelines. We present a machine learning framework for this problem and introduce a novel nonlinear optimization approach for finding the optimized weights of a sales forecasting function. Our model also optimally determines the number of historical periods to use within the framework. We present a linear alternative model to the aforementioned model and present numerical results that show the superior performance of our method. 3 - Value Of Integrated Travel Data To The Organization Productivity Pawan Chowdhary, Senior Research Engineer, IBM Research, 650 Harry Road, E3-238, San Jose, CA, 95120, United States, chowdhar@us.ibm.com, Guangjie Ren, Raphael Arar In large enterprise, travel is integral part to meet customers, attend events and to deliver services. But travel data is fragmented from planning a trip to expense submission due to the sourcing from multiple vendors at each stage. We can derive greater value by learning from the booking and spend patterns, and leverage analytics for advanced booking, to negotiate better cost with vendors, identify market with short term demand forecast, etc. which can bring ten’s of millions of cost savings. We will present our findings and analysis used to derive the savings and productivity enhancement. MA55 Music Row 3- Omni Modeling, Optimization, and Data Analytic in the Service Industry Sponsored: Service Science Sponsored Session Chair: Mohammad Sadegh Mobin, Western New England University, Springfield, MA, United States, mm337076@wne.edu Co-Chair: Zhaojun Li, Western New England University, Springfield, MA, United States, zhaojun.li@wne.edu 1 - Resource Balancing In Intermodal Freight Networks Amirali Ghahari, University of Arkansas, 4116 Bell Engineering Center, Fayetteville, AR, 72701, United States, aghahari@uark.edu, Edward A Pohl Freight transportation networks provide a system to move containers that are filled with goods from one point to another. These movements are the main source of profit for companies. When each node in a network does not have equal number of incoming and outgoing containers, some nodes will have surpluses while shortages occur at others. This fact causes accumulation of containers at a few nodes in the network and shortages at others which would shut down the transportation network. To resolve this, operators should perform rebalancing moves. This research examines the planning problem to balance resources in an intermodal transportation network for one of the major transportation companies in the US.

4 - Real-time Data In Humanitarian Response Kezban Yagci Sokat, Northwestern University, kezban.yagcisokat@u.northwestern.edu, Irina Dolinskaya, Karen Smilowitz State of the art humanitarian logistics models have been developed over the past decades. Most of these models assume availability of data. We study the impact of granularity in real time data on the humanitarian logistics models. We show that in the limited data environment higher granularity might lead better results. MA53 Music Row 1- Omni Decision Analytics for Technology Management Sponsored: Technology, Innovation Management & Entrepreneurship Sponsored Session Chair: Tugrul Daim, Professor, Portland State University, Engineering and Technology Management Department, P.O. Box 751, Portland, OR, 972070751, United States, ji2td@pdx.edu 1 - GPS For Innovation Jianxi Luo, Singapore University of Technology & Design, luo@sutd.edu.sg Engineers, firms or governments continually explore innovation opportunities and roadmaps. However, related activities and decisions are traditionally based on intuition or experiences. InnoGPS is developed to provide scientifically-grounded and data-driven support for decisions regarding innovation directions. It integrates an empirical network map of technologies that represent the total technology space, and various map-based functions that allow users to navigate through the technology space, locate themselves, explore technologies within and across neighborhoods, and identify capability-building paths. InnoGPS is a “GPS for Innovation” in the technology space. 2 - Integrating Bibliometrics And Social Network Analysis For Identifying Knowledge Sources Tugrul U Daim, Portland State University, ji2td@pdx.edu, Edwin Garces At an era when technologies are developing rapidly, decision making becomes even more challenging. However data analytics have shown that data can be used effectively to help decision making in such environments. Several management strategies for technological innovations require expert judgments and thus making the expert identification very crucial. This paper integrates SNA and Bibliometric Analysis to determine the lead authors and their network. The main objective of this paper is to present cases from the power sector where this method was used to identify experts for applications such as technology roadmapping or forecasting. 3 - Evaluating Research Centers: Case Of NSF’s I/URUC Program Elizabeth Gibson, Portland State University, 14396 SW Pennywort Ter, Tigard, OR, 97224, United States, elgibson@pdx.edu This research is focused on gaining deeper insights into US National Science Foundation (NSF) science and engineering research center challenges and motivated to develop a method that effectively measures the performance of these organizations. While research has addressed organizational performance at the micro, or single-actor level for universities or companies and at the regional or national macro level, the middle level where the NSF centers reside is largely missing. The bulk of the cooperative research center studies use either case-based methods or bibliometric data to measure traditional research outputs. Many are excellent studies; however, they only focus on a piece of the performance measurement problem. There is a need for more research to understand how to measure performance and compare performance of cooperative research centers formed in a triple-helix type partnership involving government, industry and academia. 4 - Design Support Of Salient Research Project By Integrated Approach Of Text And Citation Analysis Yuya Kajikawa, Tokyo Institute of Technology, kajikawa@mot.titech.ac.jp Bibliometrics has been a powerful tool to comprehend the current status and to analyze R&D trends but most of approach is descriptive. We proposed alternative approach to design salient research project by integrating citation analysis with text analysis. Explicit research cluster is extracted by citation relationships and implicit potential ones are by text analysis. This approach can help to find neglected opportunities between different research domains. This approach can also visualize plausible path how academic can contribute to development of industrial technology and to solve social issues. Efficiency and effectiveness of the approach are demonstrated in case studies.

139