Fig. 3. Classification tree for predicting acceptance/rejection of the 978 papers submitted to the Korean Journal of Anesthesiology from 2012 to 2015. Classification is attempted with 3 predetermined variables such as the type of the paper, the submission year, and the similarity index. A recursive partitioning model involves the similarity index, ‘simil’ in four splits. Each node represents the expected decision, its probability in parenthesis, and the number of papers in the node. Each arrow denotes a branch of the split, in which a condition is represented. To facilitate easy interpretation, a reference node is drawn in a box fenced with a broken line.