What can you learn about wildfires when you provide a room full of analysts with 7 years of US wildfire data and the tools they need to analyze it? A lot.
At a recent data dive, we split 35 data scientists into 9 teams, provided multiple data sets containing information about wildfires around the United States from 2010-2016, and gave them almost 8 hours to make predictions about future wildfires.
Here are just three of the many interesting predictions they made:
- On average, Louisiana is the state predicted to have the highest total damage due to wildfires.
- The month of August is predicted to have the largest amount of total damage on average due to wildfires.
- In California, the month of June is predicted to be the costliest for damages due to wildfires.
But the data dive teams took their analyses beyond fact finding. The winning team developed a model that could help state agencies predict the amount of resources needed each month to combat wildfires. The second-place team discovered wildfire trends using text analytics to explore text and voice data from wildfire reports.
What could you discover with this same data? And how would you do it? Most of the data was sourced from US government Web sites, including this NOAA data set.
Register for an upcoming SAS hackathon and read more hackathon stories.
