Every organization, whether public or private sector, is working hard to use its resources efficiently, but nowhere is the pressure more intense than in local government agencies. Property tax income has declined in recent years, and staffing shortages are pervasive. Local government has responded by leveraging KM solutions to support decision-making, allocate resources and improve efficiency.
Arlington County, Va., has a population of more than 200,000, including about 20,000 students in the public school system. Enrollment has increased nearly 20 percent over the past five years, and the school system is projected to reach capacity by 2013. To provide enough seats for students, expansion will be necessary, which could be accomplished through a number of routes, including new schools and portable classrooms.
Like virtually all state and local governments, Arlington County must make tough decisions about how to use its resources. The county was seeking an optimal solution for accommodating its growing number of students, but also wanted to enhance its decision-making process. Alison Denton, director of facilities planning for the Arlington Public Schools, says, "Our residents place a high value on community process and engagement in decision making. We were looking for a way to provide a structured and transparent decision-making process to show the logic that led to the eventual decisions."
Explaining values and priorities
The school board decided to use Decision Lens 3, a decision support software product. Decision Lens provides a systematic way of comparing decision criteria and values leading up to the final choices, in the context of the overall goal and strategic objectives involved. "Before talking about any specific projects, we had public meetings in which we talked about values and priorities," Denton explains. "We were able to show, through the Decision Lens application, what factors we had considered, and explain why we prioritized them as we did."
The criteria were set by the school board, because that group was ultimately the decision-making body, but the format and process were engaging, and by the time that the final capital improvement plan was presented, only two people attended the board meeting to make comments on the plan. "Everyone had seen the process, and we had a high level of buy-in, in a way that we had not achieved previously," Denton says. By managing the decision-making process in a well-documented and structured way, Decision Lens allowed stakeholders to see that while many factors were considered, including impact on open space, parking and transportation, the main goal was to provide more seats for students in the most economical way.
The school system also might use Decision Lens for other decision processes. "As our population increases, we will eventually need to reconsider school boundaries," Denton explains. Always an emotional issue, the modification of boundaries could be made more tractable by demonstrating to citizens that numerous factors were considered and addressed in an equitable way. "We would like to use Decision Lens to talk about these criteria and prioritize them upfront, to create a continuous feedback loop between the community, a group of advisers and the school board," she adds.
Decision Lens has potential to support other decision-making processes, including such tasks as HVAC maintenance and even selection of classroom textbooks, according to Denton. "In each setting, Decision Lens would require a champion, but does make available a framework for clarifying the thought process of our decision makers, and a way to communicate both within and outside of our organization," she says.
If the users decide that the priority of a certain factor has changed, they can see how that would affect the allocation of resources. "In one situation relating to a transportation agency," says Daniel Saaty, CTO of Decision Lens, "there was a major accident, and it was very easy to re-prioritize projects based on moving ‘safety' to the top." Decision Lens incorporates human judgment into decision making in situations where value is not easily quantifiable. "We are doing ‘people mining' rather than data mining," explains Saaty, "to make those judgments explicit."
KM on the crime beat
In Santa Cruz, Calif., the police force has successfully reduced crime by nearly 20 percent in the past year, aided in part by a software product called PredPol, which predicts areas where crime is most likely to occur. "We knew that getting more officers on staff was not a realistic option," says Zach Friend, crime analyst for the Santa Cruz Police Department (SCPD). "We were looking for a way to allocate our resources more efficiently."
In exploring software solutions that might help support resource allocation decisions, Friend came across a research report on a project that had brought together an interdisciplinary team, including an anthropologist and mathematicians, to develop models that reflected patterns of crime. "The software had been tested on a sample of data but not in an operational setting," Friend says. "We partnered with the research group to find out whether the software could be effective in a real-world setting, using the police reports in our records management system."
Before moving forward with the effort, the SCPD tested the system by providing five years of information to PredPol and challenging it to predict a sixth year, for which the department already had information. "The software predicted a significant percentage of crime for that year, so we decided to incorporate it into our operations," Friend adds.
The SCPD spent about six months working with researchers to determine what the output of the analyses should look like, and launched the PredPol system in fall 2011. The software tracks residential and commercial burglaries, auto break-ins and auto theft.
Likely "hot spots"
Before each shift starts, a patrol sergeant or supervisor logs onto the system and the screen shows 15 "hot spot" locations outlined in a red box. A map is printed out that officers take with them on their patrol. "We have 13 square miles to cover," Friend says. "The software identifies the zones where we are most likely to have problems and therefore lets us use our resources better." Police reports are automatically run though the PredPol algorithm every hour to keep the system updated.
In any given hour, an officer has approximately 15 minutes that is not dedicated to an activity such as responding to a 911 call or conducting pre-scheduled patrols. During that time, the officers move into the red box areas. In several dozen cases, that process has produced arrests by officers who were in the right place at the right time. However, the ultimate goal is to prevent crimes from occurring in the first place. "The arrests provided good feedback that the system was working," Friend explains, "but PredPol also reduced crime through prevention."