Problem being addressed
As the climate crisis deepens wildfires have become more prevalent. How can human induced wild fires be predicted and prevented?
A number of environment and human factors play a role in the creation of and the spread of wildfires. These range from poor land management practices to drought conditions. In order to quantify and predict where the risk of a wildfire is high or low, this research considers this risk indicator as an dependent variable with a number of causal factors as independent variables. Based on data taken from peninsula Spain over a number of years, various machine learning models including regression, random forests and support vector machines are used to mimic the relationship between the variables in the dataset. Random forests were about 75% accurate in predicting the risk of fire.
Advantages of this solution
Wildfire is a serious threat to human and animal life. This research is perhaps an early step into developing a predictive methodology for wildfire monitoring.
Solution originally applied in these industries
Possible New Application of the Work
One aspect of this research was to predict water levels in plant material. This could be reapplied to the forestry industry for monitoring purposes and for the design of firebreaks.
Viral epidemics spread in similar patterns to fires moving from individual to individual until the virus is contained or the hosts are saturated. This research could be used in the healthcare sector to better understand and predict epidemic behavior.
Viral marketing is an area where users move a marketing message among themselves at a rapid rate, and there is very little for the marketer to do with regard to advertising. This is when a message spreads like wildfire. Perhaps an AI could be designed to predict and understand when a message will go viral?
Source DOI: #############
Source URL: #############