Heuristic Search and Information Visualization Methods for School Redistricting

We describe an application of AI search and information visualization techniques to the problem of school redistricting, in which students are assigned to home schools within a county or school district. This is a multicriteria optimization problem in which competing objectives must be considered, such as school capacity, busing costs, and socioeconomic distribution. Because of the complexity of the decision-making problem, tools are needed to help end users generate, evaluate, and compare alternative school assignment plans. A key goal of our research is to aid users in finding multiple qualitatively different redistricting plans that represent different tradeoffs in the decision space. We present heuristic search methods that can be used to find a set of qualitatively different plans, and give empirical results of these search methods on population data from the school district of Howard County, Maryland. We show the resulting plans using novel visualization methods that we have developed for summarizing and comparing alternative plans. Motivation and Overview This research focuses on developing decision support tools for the problem of school redistricting. In this domain, the goal is to assign the students from each geographic region (neighborhood or planning polygon) in a county or school district to a home school at each level (elementary, middle, and high school). We are working with the Howard County, Maryland, school system to develop tools that will aid in generating, evaluating, and comparing alternative school assignment plans. Related applications include emergency response planning, urban planning and zoning, robot exploration planning, and political redistricting. The school assignment plan should ideally satisfy a number of different goals, such as meeting school capacities, balancing socioeconomic and test score distributions at the schools, minimizing busing costs, and allowing students in the “walk area” of a school to attend that home school. Since these objectives are often at odds with each other, finding the best plan is a complex multicriteria optimization problem. It is also often desirable to create several alternative plans for consideration; these plans should be qualitatively different— that is, they should represent different tradeoffs among the Copyright c © 2006, American Association for Artificial Intelligence (www.aaai.org). All rights reserved. evaluation criteria. Finally, because of the complexity of the problem, it is difficult for users to fully understand these tradeoffs. Therefore, developing effective visualizations is an important challenge. The contributions of our work are (1) a computational formulation of the school redistricting problem as a multicriteria optimization problem; (2) novel heuristic local search techniques for generating high-quality, diverse (i.e., qualitatively different) plans; (3) visualization methods1 for comparing alternative plans; and (4) empirical results demonstrating the effectiveness of our search methods on actual Howard County school data. The rest of this paper is organized as follows. We first describe the current redistricting process in Howard County, and present some example plan visualizations that we have developed. Next, we describe the search methods and present empirical results comparing manually and automatically generated plans in terms of plan quality and diversity. Finally, we summarize related work, then present our future work and conclusions. Redistricting Process The Howard County Public School System serves a rapidly growing county in suburbanMaryland. The pace of development and population growth has necessitated the opening of 25 new schools in the last 14 years, turning the adjustment of school attendance areas into an almost annual event. Under the current process, candidate plans and feasibility studies are generated manually2 by school system staff. These plans are evaluated and refined by a committee of citizens, then presented at regional meetings for public comment. A small set of candidate plans is forwarded to the Superintendent, who presents two or three recommended alternatives to the Board of Education. The Board has final decision-making authority, and will typically select one of the recommended plans, sometimes making minor modifications in response to concerns raised by parent groups or staff. Note that this These visualization methods are summarized here; they are described in detail in an earlier publication (Shanbhag, Rheingans, & desJardins 2005). Map-based tools are used to show the proposed school districts, and a set of spreadsheets is used to generate evaluation data. No other decision support tools are used in the current process.