“…Following a genomic prediction approach previously employed for sugarcane (Aono et al, 2020;Pimenta et al, 2021), ML algorithms coupled with FS were used to predict the attribution of genotypes to two groups: those that presented mosaic symptoms at any block or year (susceptible) and those that did not present symptoms in any case (resistant). Eight ML algorithms implemented in the scikit-learn Python 3 module (Pedregosa et al, 2011) were tested: adaptive boosting (AB) (Freund and Schapire, 1997), decision tree (DT) (Quinlan, 1986), Gaussian naive Bayes (GNB) (Friedman et al, 1997), Gaussian process (GP) (Rasmussen, 2003), K-nearest neighbor (KNN) (Cover and Hart, 1967), multilayer perceptron (MLP) neural network (Popescu et al, 2009), random forest (RF) (Breiman, 2001) and support vector machine (SVM) (Cristianini and Shawe-Taylor, 2000).…”