Grafting is the process of adding nodes to inferred decision trees to improve the predictive accuracy. A decision tree is a graphical model that is used as a support tool for decision process.