“…In lacustrine ecosystems, the pairing of longterm and or high-frequency databases and a random forest (RF) model, seems to be a good alternative for hydro-ecological models for predicting the phytoplankton biomass and bloom (Breiman, 2001;Thomas et al, 2018;Yajima and Derot, 2018). Moreover, some clustering algorithms based on machine learning models such as the Kmeans method, enable the creation of classes to address environmental problems (Derot et al, 2020;Hartigan and Wong, 1979;Rousseeuw et al, 2014;Solidoro et al, 2007). Furthermore, the predictive performance of an RF model can be improved by utilizing it in conjunction with a K-means model (Kwon and Park, 2016;Liu and Sun, 2019).…”