Study on Most Important Factor and Most Vulnerable Location for a Forest Fire Case Using Various Machine Learning Techniques

摘要

Forest Fire is a nature disaster that can cause a series of consequences and impacts. This paper studies the Most Important Factor (MIF) and Most Vulnerable Location (MVL) for a forest fire case in Montesinho Natural Park using various machine learning techniques. We applied four algorithms to analyse the data and derive MIF and MVL. Hence, we evaluated experiment results to explore the relationships between the algorithms. We conclude that MIF can be achieved using both regression and classification approach, one using the corresponding weight value and one using the variable of the split cut. For MVL, it can be derived from the centroid of clusters and a heuristically counting of the frequent location point.