Decision tree

A decision tree shows a connected hierarchy of boxes to represent the values of records. The records consist of decisions and their possible outcomes, and highlight key insights about the decision tree.

Records are divided into groups, which are called nodes. Each node contains records that are statistically similar to each other regarding the target field.

Nodes can be the root node, internal nodes, or leaf nodes. When you click an internal node, its leaf nodes are hidden which helps you to focus on specific records. You can click a leaf node to view the path from the root node.

You might notice a mini map in the visualization, if there are many nodes.

Decision tree minimap

Each branch in a decision tree corresponds to a decision rule. A decision rule predicts an outcome in the target field. The decision rules help you determine which conditions are likely to result in a specific outcome.

Decision trees have a single target. If the target field of the decision tree is continuous, then the key insight indicators show significantly high or low groups. You can see that the color of the node is based on the average of the target for the measure. The higher the average value of the target for a node, the darker the color. If the target field of the decision tree is categorical, then the key insight is the mode of the node. The mode of the node is the most frequently occurring category (or categories) of the target field within the group.

To edit or add key drivers, click the More icon More icon on the Target field.

From the Nodes drop-down list, you can choose any of the following options to view certain nodes:
  • All: A full tree is shown with all values.
  • Top 5: The top five highest target values are shown.
  • Bottom 5: The five lowest target values are shown.

If you have a categorical measure, then after choosing the nodes Top 5 or Bottom 5 you can select a category for which you want to see choose nodes from the Target category drop-down list.

Insights are different depending on the type of target. If you are predicting a continuous measure, then the decision tree shows the average value of the target within the node given how far down the node is in the path.

For target fields that are a continuous measure, all nodes have the same color but with different darkness.

For target fields that are categorical, the decision tree has different colors in each node that represents each category.

The following image shows two examples of a decision tree visualization. The example that is on the left side of the image shows a decision tree with continuous measures. The example that is on the right side of the image shows a decision tree with categorical measures.

These examples shows a decision tree with continuous measure that is on the left side of the image and a decision tree with categorical measure that is on the right side of the image.

To re-create these visualizations, complete the following steps:
  1. From a dashboard, click the Visualizations icon and in the System tab of the Visualizations pane, select Decision tree.
  2. Click the Select a source button in the Data pane, and in the Select a source window go to Samples/By industry/Insurance/Data. Select the Customer analysis data module and click Add.
  3. Drag the Monthly Premium Auto onto the Target field.
  4. To turn off the mini map in the visualization, click Properties and in the Chart section of the Visualization pane, click Show navigation mini map to turn off this map.
  5. Click the Visualizations icon and in the Visualizations > System tab, click Decision tree.
  6. Drag the Expiry Month data set onto the Target field in the Fields pane for the second decision tree visualization.

Samples dashboard

An example of a decision tree visualization is available in the Telco churn dashboard, on the Tenure tab.

Location
Team content/Samples/By industry/Telecommunications/Dashboards/