9:10am-10:00am
Creating Structure and Order from Chaos
Tobin J. Lehman, Services Scientist, Almaden Services Research Group, IBM Almaden Research Center
In the business of strategic outsourcing, the company supplying the computing technology and expertise (i.e. the selling company) must satisfy the requested requirements specified by the receiving company. The list of provided services in a particular customer engagement (e.g. help desk support, network data support) must match the customer’s needs, and yet they must also match the services that are listed in the selling company’s service catalog. The potential mismatch in service descriptions can result in chaos for the selling company, as the mechanism for tracking internal cost and external price is gone. In this presentation, Lehman will show the IBM solution to the service mismatch problem described above. He will describe both the methodology used in the solution and the taxonomy used to label components of the actual delivered services. The result is a consistent mechanism for characterizing services and determining the appropriate price that goes with a service, based on actual experience. Co-author: Greg Eyres.
10:30am-11:20am
Dynamic Cost Reduction through Process Improvement in Assembly Networks
Fernando G. Bernstein, Associate Professor of Operations Management, Fuqua School of Business, Duke University.
Manufacturers and their suppliers invest in process improvement activities. As a result, production costs decrease over the life cycle of a product. Some of this cost reduction is reflected in the suppliers selling prices to the manufacturers. This presentation explores how market conditions, supply network structure, and procurement arrangements affect the magnitude and timing of investments in cost reduction, as well as all the firms’ profit performance. Bernstein will consider two approaches to procurement commonly used in practice. Under cost contingent contracts, the manufacturer adjusts the prices it pays to its suppliers each year in response to their achieved cost reductions. Target-price contracts specify a time-path of target cost reductions, and suppliers keep any amount of cost reduction beyond those stipulated in the contract. The work of Bernstein and his colleagues suggests that implementing an appropriately designed target-price contract may benefit all firms. Based on their findings, Bernstein will propose a number of insights regarding the effective design of procurement contracts.
11:30am-12:20pm
Optimizing Integrated Direct Marketing Campaigns
Dmitry Krass, Professor of Operations Management and Statistics, Rotman School of Management, University of Toronto, and Senior Partner, R&D, Custometrics, Inc.
A typical direct marketing (DM) campaign is concerned with identifying the best targets for a given product and then contacting these targets through selected DM vehicles. Predictive statistical models are often used to rank the list of prospective targets (customers or geographical regions). However, this traditional paradigm assumes each DM campaign is planned and executed in isolation. In reality, many companies operate in environments that are multi-product (i.e., several different products or offers may be marketed simultaneously) and multi-period (i.e., available budget has to be allocated over a number of upcoming DM campaigns). In this session, Krass will describe an integrated DM campaign management system that addresses these issues, modeling issues that arise, and his team’s experiences with the implementation of the system. He will also discuss several future challenges, such as integrating mass and direct channels.
3:10pm-4:00pm
Olefins Furnace Feed Scheduling Decision-Support System Using Optimization
Vijay Hanagandi, Chief Technical Officer, Applied Optimization, LLC
The scale of the operations and the material logistics requirements in Olefins manufacturing requires careful planning and scheduling of operations. This is compounded when there are independent plants situated miles apart that have to be coordinated to optimize company-wide metrics. In this project, Hanagandi and his colleagues used AIMMS to develop a decision-support system to manage multi-site logistics of such a manufacturing business. This work is focused on using non-linear programming to facilitate targeting furnace operating conditions in order to derive furnace feedstock flowrates. The relationship between plant operating conditions and the feedstock flowrates needed is complex and at times non-linear. The IT architecture needs constrained the team to design a thin-client solution with a common model and optimization engine on the server. Hanagandi will also discuss the IT aspects of the deployment, along with the mathematical problem solution. Co-authors: Sheila Schmitt, Paul Leggett, LyondellBasell Industries.
4:10pm-5:00pm
Policy Evaluation by Case Matching: Bridging Statistics and Optimization to Uncover Causal Mechanisms
Yanni Papadakis, Vice President, Management Science, Business Intelligence, Merrill Lynch & Co.
The exciting method of Case-Matching is increasingly finding applications in high power analytics. Through his experience in financial services, Papadakis will present important examples of matching exercises that clarify the uses and abuses of this family of methods. The audience needs no special prior knowledge in Case-Matching; the method will be introduced in the presentation. The emphasis of this talk is on implementation, a topic near and dear to the heart of every analytics professional. Two practical problems will be analyzed in depth: having numerous dimensions to match on, and picking matches out of millions of candidates. Analytics practitioners with previous experience in Case-Matching will be pleasantly surprised to find how at Merrill Lynch optimization tools have added efficiency and flexibility to statistical analysis.
|
