Predicting the future with better data visualization
Above: Ross Maciejewski is exploring spatiotemporal data to find patterns and identify anomalies that help forecast what might occur in the future. Photographer: Jessica Hochreiter/ASU
Imagine you operate an amusement park and you want to ensure your park is optimized for both guest satisfaction and operational efficiency. One way to do it is by tracking how visitors move throughout the park. Monitoring that data will allow you to improve the park layout, manage ride wait times and generally improve operations.
That type of data is no longer limited to video games such as Roller Coaster Tycoon, it is part of the research Ross Maciejewski, an associate professor of computer science in the Ira A. Fulton Schools of Engineering at Arizona State University, is conducting as he focuses his work on how to enable the exploration and communication of data.
Specifically, he looks at data, such as crime statistics, census data and climate data, which all encompass a specific geographic location and time.
“What we’re trying to do is develop new solutions for exploring this large spatiotemporal data to find patterns and identify anomalies,” Maciejewski says. “Our goal is to help analysts build models that can forecast what might occur in the future. This allows for better planning and management.”
Maciejewski, the director of the Center for Accelerating Operational Efficiency, is also expanding this work as some of his research has looked at how people might walk through a city and how to organize landscaping to provide better shade.
In his lab, Maciejewski is looking at ways to help designers better convey their messages in maps like those seen during election coverage that are known as choropleth maps.
“We’re looking at choropleth maps that put data into discrete buckets; however, the choice of how those buckets are decided can be done in numerous ways,” explains Maciejewski. “We’ve been developing tools to try and help designers understand how well their design is actually conveying the real data.”
Yifan Zhang, a former doctoral student and member of Maciejewski’s lab, worked to explore how various map classification schemes could be manipulated to create different messaging.
For example, imagine an election map of the United States split into two colors, red and blue. Election results for each state will give each color a value between 0 and 100 with the color of each state determined by their given value.
That methodology may give you an answer, but it won’t give you the full story of the data. If you split red and blue at above and below 50, then states with a value of 49 get placed in blue even though this value is very close to being in the red.
“We want to see how much these sorts of issues impact the map messaging and provide map designers with an awareness that this is occurring,” explains Maciejewski.
The beginning of a research journey
This work is all related to the National Science Foundation’s CAREER Award Maciejewski was awarded in 2014.
“The main goals of our research when we first set out was, how can we effectively explore such space-time data in order to enhance knowledge discovery and dissemination,” says Maciejewski.
The team produced a solid body of literature demonstrating challenges and potential solutions in this area and identifying areas where more work is needed. This includes 10 journal publications with five of those appearing in IEEE Transactions on Visualization and Computer Graphics, one in the Annals of the American Association of Geographers and seven conference publications.
“We had envisioned particular domains that we were going to explore such as crime, climate and health,” says Maciejewski. “However, given the abundance of these types of data, we were able to really hit new areas, including movement data like taxi data and amusement park visitors as well as global trade data. This allowed us to see how well things might generalize and really strengthened our findings.”
Looking back to move forward
For Maciejewski, the NSF CAREER Award represented his first significant research recognition.
“It served as a really important mechanism for me to pursue some of my interdisciplinary collaborations,” says Maciejewski. “For visualization, we need data and domain problems.”
Working with climate scientists, urban planners, criminologists and other experts allowed him to think about different problems and domains he could pursue.
“It helped me develop other research ideas that could spin out from this work,” says Maciejewski. “I’m very grateful for this award and the support that I had at ASU for doing interdisciplinary research.”
Developing future researchers
The NSF CAREER Award requires an education component to the recipient’s project. For Maciejewski, this involved developing an entire college course.
“I developed the class known as Introduction to Undergraduate Research,” says Maciejewski. “This course is open to any student in the Fulton Schools and is primarily targeted to freshmen and sophomores who want to learn about the research that goes on at ASU and how to get involved.”
At least 50 percent of the students engaged in the class went on to conduct undergraduate research in one form or another according to surveys done by Maciejewski. Several students from the course even joined his lab, with Scott Freitas, Rolando Garcia and Alexandra Porter each receiving NSF Graduate Research Fellowships to continue their studies in graduate school.
“It’s been great to see students I’ve worked with as undergrads at ASU go on to Berkeley (Garcia), Stanford (Porter) and Georgia Tech (Freitas), to name a few,” says Maciejewski. “I’m excited to see their PhD research.”