The x input is the first training data item. The predicted y value is 0.4909, which is quite close to the actual y value of 0.4840. Because of the way decision trees for regression are created/trained ...