You can think of these as the roots of our tree. The roots of our tree first include regression. In our LNOB trees we are creating regression trees that are calculating the average access rate at every node of the tree. Second, our trees are Binary, which means that each branch splits into a maximum of 2. Third, we must decide how our trees are being split. In our case we use the Analysis of Variance, ANOVA for short. ANOVA uses the sum of squares method for splitting the data to identify the highest variance. And lastly, we define and use a splitting criterion  to check to see if a node can be split any further. These elements together create the roots of our LNOB trees.