Machine Learning Classification Algorithms for Predicting Karenia brevis Blooms on the West Florida Shelf

Harmful algal blooms (HABs), events that kill fish, impact human health in multiple ways, and contaminate water supplies, have increased frequency, magnitude, impacts numerous marine freshwaters around the world. Blooms of toxic dinoflagellate Karenia brevis resulted thousands tons dead deaths to many other organisms, respiratory-related hospitalizations, tens hundreds millions dollars economic damage along West Florida coast recent years. Four types machine learning algorithms, Support Vector Machine (SVM), Relevance (RVM), Naïve Bayes classifier (NB), Artificial Neural Network (ANN), were developed compared their ability predict these blooms. Comparing 21 year monitoring dataset K. abundance, RVM NB found better skills bloom prediction than two approaches. The importance upwelling-favorable northerly winds increasing probability, onshore westerly preventing from dispersing offshore, quantified using RVM, all models used explore large river flows nutrients they supply regulating These provide new tools for management devastating