Project titles from previous quarters
Title: An Outreach Index for Washington State Opportunity Scholarship
Authors: Andrew Chipperfield, Ian McDermott, and Zhanxi Shi
Abstract: The Washington State Opportunity Scholarship helps provide low-income students with a chance to pursue higher education in science, technology, engineering, math, or health care. In recent years, the scholarship has received more money to distribute and seeks to receive more applicants for their scholarship through targeted promotion at high schools. With a staff of only ten people, they need to choose where they go to promote wisely in order to increase the number of qualifying applicants. As such, our task is to find a way of choosing high schools that Washington State Opportunity Scholarship can target and expect the most eligible applicants from their efforts. We based our model on available information of the schools about the number of students that meet the eligibility criteria for the scholarship. With this model, we can retrieve a ranked list of the schools based on the eligibility of their students, which should let Washington State Opportunity Scholarship staff focus their efforts on schools with more eligible students.
Abstract:
United Way of King County’s Free Tax Campaign o↵ers free tax preparation for low income residents in the greater Seattle area. For many years, United Way has been using the same work flow model, but is now considering switching to a new strategy that could potentially serve more clients. We created Monte Carlo simulations to determine how many clients are served for a given number of volunteers and site managers scheduled. By comparing outputs for many runs through both simulations, we concluded that it will be beneficial for United Way to change to the proposed new strategy. As the Free Tax Campaign grows, the new service model will serve more clients than the previous model. These simulations can also be used to determine how many volunteers and site managers should be scheduled at a tax site in order to maximize the people the Free Tax Campaign can serve.
Abstract: Northwest Harvest is a nonprofit organization dedicated to fighting hunger throughout Washington State. They allocate donated foods and resources to people in need from three warehouses located in Kent, Spokane, and Yakima. In this paper, we mathematically model their current distributions and solve for an optimal allocation of all types of food stored in each warehouse to the counties they serve by implementing the mixed integer linear programming solver in SageMathCloud. The results of the optimal model will be used as a baseline measure for Northwest Harvest to evaluate the impact of their current distribution strategy. Furthermore, it allows for more informed decisions in how they wish to allocate food to reduce statewide hunger more effectively. Our objective is to maximize the impact of the distributed food on the food insecure population in Washington State. The results show that Northwest Harvest’s current distribution method is not far from optimal and with supplementary analysis, we are able to identify counties that would most benefit from additional support.
Abstract: The manager at Little Sheep Mongolian Hot Pot wishes to find a more standardized food ordering schedule. Heuristic approaches and waiter experience are not always reliable for determining portion sizes, especially for large table sizes. A model of consumer demographics and eating habit was built through collected data. Different approximations on the ideal schedule were then tested using Monte Carlo simulations until goals with regards to minimizing waste while meeting customer satisfaction were met.
Abstract: The FirstYear Programs (FYP) office at the University of Washington has the task of scheduling and pairing Freshman Interest Groups with student leaders, a process which would take their team hours to complete due to the large number of students’ class schedules to work around, student preferences, and administrative constraints. We addressed this issue by using combinatorial optimization algorithms to build a program that would produce a feasible matching that the FYP offices can use every year.
Abstract: The Pacific Science Center (PSC) is a nonprofit organization that aims to "inspire a lifelong interest in science, math and technology by engaging diverse communities through interactive and innovative exhibits and programs" (PSC). Located in Seattle, Washington, PSC has roughly a million customers per year, and often more than 3000 customers per day (PSC). Tickets for various PSC events can be purchased at two entrances. Special events and seasonal variances affect the traffic flow into PSC. The number of cashiers at each entrance at different times during the day should optimally vary to account for these changes, as too many cashiers is a waste of resources, but not enough cashiers results in poor customer service. Using Monte Carlo simulation, we studied the changes in customer wait time, customer line length, and cashier idle time for different quantities of cashiers on duty. By running simulations on a comprehensive range of customer arrival rates, we determined the minimum number of cashiers to place at each entrance in order to keep the probability that a customer waits more than three minutes under 5%. We also looked at the efficiency of automated ticket machines.
Abstract: Engineers Without Borders (EWB) is a nonprofit organization that does engineering work to improve lives of people and communities in the most remote places around the world. EWB-UW Chapter has been working with a community in the Andes Mountains of Bolivia, improving a 15-mile stretch of steep, mountainous road that provides the community their only access to the outside world. The problem that the EWB team faces is scheduling approximately 35 workers to work on 54 different jobs (digging ditches, building walls, etc.) during the 20 days of the year in which EWB is there. Because of this limited window, EWB-UW desired to optimize their scheduling so that the maximum amount of work was completed and the more important jobs were done first. The method implemented to try and solve this is a branch-and-bound algorithm to generate partitions of the initial set of jobs. From a given dataset, the program will attempt to enumerate every valid partition of the listed worksites and find an optimal one that minimizes the intra-cluster variance in job importance, and attempts to make every cluster require an integer number of days to complete.