XLSTAT: Customize a Decision Tree in Excel

 

Dataset for Example

See attached.

For this example, we will start with an existing decision tree and to configure its display.

 

Setting Up

Start with the existing tree from the dataset. In the decision tree toolbox, click Remove Grid to enhance the display.

   

The tree displays horizontally, with branch blocks above node blocks. To adjust the layout, right-click the first tree block on the left and select XLDTREE > Open the settings dialog box for the selected tree.

 

In the Options tab, you can adjust the display of the tree and branch/node blocks with four available styles. After exploring each option, choose the one that best suits your needs. This setting can be changed anytime, and the tree will update once you click OK.

• Tree display: Horizontal, Branch/Node block: Top/Bottom: This is the default display.

• Tree display: Horizontal, Branch/Node block: Left/Right: The blocks will be arranged horizontally, with branch and node blocks aligned left and right.

• Tree display: Vertical, Branch/Node block: Top/Bottom: The tree is displayed vertically, with branch blocks positioned above the node blocks.

• Tree display: Vertical, Branch/Node block: Left/Right: The tree is displayed vertically, with branch and node blocks aligned from left to right.

In the Useful Information tab, you can select and label the information displayed for each node. For example:

• Position 1: Expected gain (label EG) – based on the expected gain of child nodes.

• Position 2: Gain (label G) – the sum of cost and profit from parent nodes.

• Position 3: Cost (label C) – the sum of all parent node costs.

• Position 4: Profit (label P) – the sum of all parent node profits.

Here is the tree displayed vertically, with branch blocks positioned above the node blocks:

 

Additional Info

In XLSTAT, various calculation modes are available to analyze a decision tree. You can also display the optimal path based on the objective (maximize or minimize gain), either for the entire tree or from a specific node.