Big Data Predicts Wildfires

Big Data Predicts Wildfires

When a team of University of Oregon undergraduate students first started working on a data analysis problem to understand the conditions, causes, and effects of wildfires, little did they know that their model would accurately predict the wildfires that blazed across the western United States this past summer.

The students from the UO's Lundquist College of Business used analytics skills and SAS software to create a methodology for accurately predicting wildfire breakouts for their entry into SAS's 2017 Analytics Experience Shootout Competition. Their work earned them the number one spot in the competition, where they beat out 56 teams from other universities, many of which included master and PhD students. It was the second year in a row that a team from the Lundquist College won the competition.

Carmen Huang, Yue Jin, Haiming Kuang, and Reisa Widjaja, worked from February to June, using the SAS platform and analytics skills to ultimately produce an 80-page report.

“It's not a toy project—it's a real one," said UO associate professor of operations and business analytics Zhibin (Ben) Yang, who advised the team along with assistant professor of operations and business analytics Wenbo Wu.

For the first two months, the students estimate they spent 20 hours a week each working on the problem statement. After that, it was nearly full time, with the students breaking only for classes and a little sleep.

“This year's project looked to be much, much harder than all the others before," said Jin. “There were 14 data sets this year. Last year, there were only four."

Also adding to the complexity, the data sets" terminologies were not uniform.

“Different agencies have differing writing styles," explained Kuang. “Oregon may describe the fire area in acres, while California may describe it in miles. We had to account for that."

“I spent most of my time in the lab text mining and finding patterns," said Widjaja, gesturing to the college's Miles and Eleanor McKay Business Technology Center. “It was my second home."

Adding to the difficulty, Jin identified two new variables to add to the analysis just two weeks before the project was due: wildfire probably increases during weekends due to human involvement and types of wildfires are defined by region, not by state.

“It was intense, but we just had to go for it," Widjaja said.

In addition to it being the second straight analytics shootout win for the Lundquist College, it was the fourth straight win for a team associated with Assistant Professor Wu.

The team credits Wu, Yang, and their operations and business analytics coursework for providing them with a foundation on which to build their skills.

“Wenbo Wu and Ben Yang were the perfect professors to teach us how to work as a team and fight for our future," Widjaja said.

“They were not just concerned about academic success, they were concerned about our lives," Huang added. “It was amazing to see our efforts pay off."

As in past years, the team was notified by SAS that it made the top three in August. In September, the group travelled to the SAS Analytics Experience 2017 conference in Washington, DC. The top-three teams are the only students granted entry to the prestigious conference for analytics professionals.

“The keynote speaker was Magic Johnson," Kuang said. “I thought if worse comes to worst, I get to see my childhood hero talk about sabermetrics in sports."

When the University of Oregon team was announced via an on-stage video, Kuang said the team was in shock. That shock quickly turned to action because the winning team must present their work to a small group of top analytics professors and industry professionals.

“The conference was such a great opportunity," said Huang. “Now we know we are good enough and want to pursue more."

Two of the students went into the competition as with a concentration in operations and business analytics. Now, all four wish to pursue the subject in either undergraduate or graduate study.

Along with national recognition, the team also earned the Lundquist College a $5,000 gift from SAS.