Fig. 3From: Predicting rice blast disease: machine learning versus process-based modelsIn the rules and tree the individual datasets cannot be distinguished/labeled, as the training set is a composite made up of all the datasets. The data in the tree and rules are distinguished by the attribute labels (e.g. relhum3, temp3, ...) and the ranges (e.g. < 23.7) but not by the datasets the records come fromBack to article page