Informs 2015

INFORMS Philadelphia – 2015

240

MD26

26-Room 403, Marriott

Analytics Maturity Model

Sponsor: INFORMS Practice

Sponsored Session

Chair: Valentina Ferretti, Politecnico of Torino, Corso Castelfidardo

30/A, Torino, Italy,

valentina.ferretti@polito.it

1 - Decision Analysis with Geographically Varying Outcomes

Jay Simon, American University, Washington, DC,

jaysimon@american.edu

This work develops theory to support decisions based on data from geographic

information systems (GIS). Preference conditions are introduced, leading to

corresponding value and utility functions over GIS data for both single-attribute

and multiple-attribute cases. These models of preferences are then applied to

example decisions based on GIS data.

2 - Application of the New Gear Geospatial MCDA Tool to

Humanitarian Assistance Site Selection Decisions

Matthew Bates, Research Engineer, US Army Corps of Engineers,

Engineer R&D Center, 696 Virginia Rd, Concord, MA, 01742,

United States of America,

Matthew.E.Bates@usace.army.mil

Patrick Doody, John Nedza, Erin Hughey, Richard Curran,

Igor Linkov, Heather Bell, Paul Kailiponi, Michelle Hamilton

Humanitarian assistance and disaster response (HADR) decisions are multifaceted,

involving many stakeholders, limited funding and competing areas of need.

Spatially explicit data are increasingly available to support these decisions at fine

scales. We introduce GEAR, a new US-Govt-developed tool for spatial multi-

criteria decision analysis that is currently being transitioned to the Pacific Disaster

Center to support the HADR community. We demonstrate its application to site

selection decisions.

3 - Key Challenges and Meta-choices in Designing Spatial

Multi-criteria Evaluations

Gilberto Montibeller, London School of Economics, Houghton

Street, London WC2A 2AE, London, United Kingdom,

G.Montibeller@lse.ac.uk

, Valentina Ferretti

Spatial multi-criteria decision analysis is being increasingly employed in

environmental decision-making and in related fields. However, there are key

challenges when designing such evaluations, which impose important meta-

choices to decision analysts, as they may lead to different contents of the

evaluation model and to distinctive outcomes of the analysis. In this paper we

provide a systematic and comprehensive discussion of these key challenges and

the associated meta-choices.

4 - Terrestrial Condition Assessment for National Forests of the

USDA Forest Service in the Continental U.S.

Keith Reynolds, Research Forester, USDA Forest Service, PNW

Research Station, 3200 SW Jefferson Way, Corvallis, OR, 97331,

United States of America,

kreynolds@fs.fed.us,

David Cleland,

Barbara Schrader, Robert Vaughan

The Terrestrial Condition Assessment of the National Forest System is using a

spatial decision support system to assess effects of uncharacteristic stressors and

disturbance agents, with an emphasis on identifying restoration opportunities at a

national scale. When outcomes were classified into categories of very good, good,

moderate, poor, and very poor terrestrial condition, corresponding percent areas

on national forests were 9.78, 46.37, 19.22, 16.75, and 7.89%, respectively.

MD28

28-Room 405, Marriott

Auctions for Spectrum

Cluster: Auctions

Invited Session

Chair: Robert Day, University of Connecticut, 2100 Hillside Road,

U-1041, Storrs, CT, 06269, United States of America,

Bob.Day@business.uconn.edu

1 - Optimal Bidding Strategies in Core Selecting Auctions

Van Vinh Nguyen, The Fuqua School of Business, Duke

University, 100 Fuqua Drive, Durham, NC,

van.vinh.nguyen@duke.edu

, Ozan Candogan, Sasa Pekec

We analyze optimal bidding strategies of a single bidder in core selecting auctions

with homogeneous items. We use robust optimization approach to formulate the

bidder’s optimization problem and show that gains from non-truthful reporting

are significant.

2 - Vickrey-based Pricing in Iterative First-price Auctions

Oleg Baranov, University of Colorado Boulder, Boulder, CO,

United States of America,

oleg.baranov@colorado.edu

Lawrence Ausubel

For achieving efficient outcomes in practical auction settings, an auction design

should use the opportunity cost pricing principle to the extent possible to promote

truthful revelation of bidder preferences, and the pricing mechanism should be

implemented via an iterative “first-price” process where all bidders are fully

informed about their current price at each iteration. For the heterogeneous

environment with substitutes, we develop an auction design that adheres to both

principles.

3 - The Use of the Clearing Target Optimization Model Within the

FCC Incentive Auction

Karla Hoffman, Systems Eng and OR Dept., George Mason

University, Fairfax, VA, 22030, United States of America,

khoffman@gmu.edu,

Brian Smith, Steven Charbonneau,

James Costa, Tony Coudert, Steve Schmidt, Rudy Sultana

We present an important component of the upcoming FCC Incentive Auction:

How to choose a “clearing target”, i.e. the maximum amount of spectrum to try to

buy back from broadcasters to sell to the broadband industry. A difficult

combinatorial optimization problem is used to determine this target. This talk will

present the formulation of this problem and our approaches to solving the

problem. We also describe how this model was used to inform policy at the FCC.

4 - Multi-option Descending Clock Auction

Tuomas Sandholm, Professor, Carnegie Mellon University, 5

000 Forbes Ave, Pittsburgh, PA, 15213, United States of America,

sandholm@cs.cmu.edu

, Tri-dung Nguyen

In descending clock auctions (DCAs), bidders (sellers) can accept or reject. Yet in

many settings, e.g., the FCC’s upcoming incentive auction, each bidder can sell

one from a set of options. We present a DCA that offers each bidder prices for the

options. We develop a Markov chain model of the dynamics of each bidder’s state,

and a method for optimizing prices to offer in each round. Experiments with real

FCC constraint data show it dramatically outperforms percentile-based price

decrements.

MD26