UC San Diego
Analyzing and Predicting Wildfires
- Author(s): Khatter, Sumedha
- Advisor(s): Altintas, Ilkay
- et al.
Wildfires can rip through any part of the world and cause a havoc. Building technologies to analyze and predicting the occurrence of the hazardous wildfires is of paramount importance. The project brings a new approach to build a system which analyzes and find areas more susceptible to wildfires using historical data of fire and landscape features. This is achieved by preparing dataset using raw data downloaded from various sources, feeding that into various machine learning models and then predicting and analyzing the results. The skewness faced by the dataset is resolved using different resampling techniques and gauged via different scores. Use of certain kind of scores for this kind of dataset and comparison of results obtained via various supervised learning models is discussed next in the project.