Performance Evaluation of Machine Learning on Forest Fire

As the technology is developing, Forest fires are one of the major environmental concerns, each year millions of hectares are destroyed over the world, causing economic and ecological damage as well as human lives. It is not always possible to cope with a fire and get to the forest in a timely manner. These disasters can cause widespread damage and destruction, so they must be identified quickly and accurately. Which is why, an early warning system can provide effective strategic data and accurate prediction results tailored for interference and fire control. This success comes with challenges and problems that are comparable with limited resources. At the moment, they have little or nothing to do with the level of risk analysis mechanisms involved. In general, the fire prediction models used today offer neither satisfaction nor accuracy. Factors that affect the prevalence of fires are the frequency and variety of fires, which are shockingly rare. This research provides an insight into the use of Machine Learning and Deep Learning models towards the occurrence of forest fire. Techniques such Random Forest, K - nearest neighbors, Artificial Neural network, Convolutional neural network has been used for prediction of forest fires. Random Forest model showed the highest forest fire prediction accuracy.