ORIE M.Eng. Projects In The Home Stretch
ORIE's Master of Engineering students have been at work on their projects for several months, and 'crunch time' is upon them. The 2011-12 projects underway in Ithaca include several that build on prior work and client relationships, as well as new projects for new clients.
Canadian National Railway
In a project that builds on past success, a team continues work on the scheduling of CN Rail work crews. Last year's project determined how that an eight hour planning horizon should be used for scheduling. This year's team is working to produce a realistic projection of both the demand for and the available supply of crew members with specific skills, within the eight hour planning horizon.
The task requires the team to compute the range of values within which the actual gap between supply and demand can be expected to fall with a high degree of confidence. They are working under the guidance of Professor Peter Jackson, who has advised the series of projects from the outset.
Ernst & Young
As a large financial services provider, Ernst & Young works especially hard to identify, acquire and serve about 1000 companies that are designated as priority accounts. Ernst & Young has asked an M.Eng. project team advised by statistics professor Paul Velleman with the assistance of ORIE professor Dawn Woodard to analyze account data, using statistical methods, to determine whether the company's strategy for these priority accounts can be improved.
For example, the team is using data to determine whether the current strategy is too focussed on clients that generate high current revenue rather than on those that have the highest potential for future income. The team includes MPS Statistics students together with ORIE students.
New this year as a project sponsor, FXI is a leader in the foam industry and holds more than 400 patents for innovative products and processes. The M.Eng. project for FXI is helping them determine best practices in scheduling the production and pouring of foam.
Doing so entails building a simulation, based on the distribution of actual order and sales data from earlier years, to evaluate different scheduling and production policies. The simulation will increase understanding the best way to react to changes in product demand. The team is also analyzing the economic impact of various sales and operating practices. Professor Jackson is the team's advisor, with the assistance of Professor Jack Muckstadt.
MITRE, a Federally Funded Research and Development Center, has sponsored a series of projects in recent years with Professor Mark Lewis as advisor. Like last year's project, which considered law enforcement applications, this year's work for MITRE deals with unmanned aerial vehicles (UAVs). If UAVs are to be integrated into the national airspace, how can they be programmed to avoid collisions with other aircraft, manned or unmanned, using rules that determine which aircraft have the right-of-way in encounters?
The situation is more complex than right-of-way rules for terrestrial vehicles, not only because of the added vertical dimension but because of uncertainty such as wind or sensor error and because diverse vehicles are likely to be in use. (The simulated flight paths at left give an idea of the complexity encountered in driving a UAV fleet). The team has built a model, based on what are known as Markov decision processes, that can be solved to give the optimal action for a UAV to take at any given time, for a given position and orientation. The model will be used in a simulation that will show the vehicle's trajectory under different scenarios.
As an ambulance service that flies patients in Ontario, Ornge must deal with schedule disruptions due to emergency requests and weather, especially in the northern part of the Canadian province. In recent years, ORIE M.Eng. teams have developed a scheduling tool that can be run every night to decide on a schedule for the next day, and then to recompute (repair) the schedule as disruptions occur.
This year's team is adding new features, redesigning the tool's user interface and streamlining the code so that Ornge dispatchers can easily find a new optimal schedule in real time at any point in the day. The team is being advised by Professor Shane Henderson; Professor David Shmoys and two PhD students are also working on aspects of the project.
Procter & Gamble
Analytics and business intelligence are playing an increasingly important role at Procter & Gamble, according to M.Eng. alumnus Jeff Goldman. The consumer products company sells hundreds of products through millions of stores and thousands of sales representatives.
P&G has asked an M.Eng. team to analyze gigabytes of data on distributor transactions and figure out which products to sell to every individual store in China so as to mazimize sales potential at each store, based on a variety of factors. Professor Huseyin Topaloglu is the team's advisor.
Following a prize-winning project for Walmart.com that helped the company forecast rates at which purchases will be returned, this year's team is focused on customer loyalty. The team is taking a detailed look at how Walmart.com compensates customers (for example with a gift card) in the event of delivery and other problems.
The team is extensively mining customer surveys and other data related to customer loyalty as a prelude to determining an optimal compensation policy. They have already found that proactive customer service actions result in higher customer loyalty, and are developing an algorithm to provide a framework for determining the most effective approach to compensation. The project is being advised by Professor Peter Frazier.
Weill Cornell Medical College Center for Sleep Medicine
New this year are two projects for the Weill Cornell Medical College in New York City. One of them is looking at obstructive sleep apnea, whose sufferers intermittently stop breathing while sleeping and then are excessively sleepy during the day. The current method used to detect obstructive sleep apnea is expensive and time consuming. It requires instrumented monitoring at a special facility throughout the night in order to provide the value of a diagnostic index called the AHI.
The M.Eng. project team, advised by Statistics Professor Martin Wells and including students in the Master of Professional Studies (MPS) in Statistics program, is attempting to determine whether the AHI can be accurately predicted by using statistical methods on other measurements that are less expensive and time-consuming to obtain.
Weill Cornell Department of Urology
An M.Eng. team advised by Professor Jack Muckstadt is working to increase the effective capacity of a 3-room cystology operating suite (one room of which is shown at right) at New York Presbyterian/Weill Cornell Medical College. During five days of on-site visits, the team has 'scrubbed up' to observe procedures and processes, interview staff, and gather data. Now they are working on a computer simulation to use in testing several ideas for improving the scheduling of the suite.
A key aspect of the project is to understand and satisfy the objectives of stakeholders ranging from patients to medical staff to room cleaners and administrators. In general the team has found stakeholders responsive, with one early recommendation implemented the day after it was proposed.