In KNIME you can color the nodes/leaves of a decision tree according to the clusters that they contain. Indeed every node or leave covers a number of patterns with different features and belonging to one of the official classes.
We can then precede the decision tree learner or predictor node with a "Color Manager" node where we assign specific colors to specific feature groups (see picture below).
The workflow depicted in the figure above was applied to the adult.data file from the UCI Data Repository. For example we can assign different colors to "Income <= 50K" and "Income >50K"; or to different jobs in the data column "work-class".
The tree view then shows the proportions of the different clusters in each node or leave.
In the figure below we colored the nodes and leaves based on "Income": red for "Income <=50K" and blue for "Income >50K". "Income" was also the data column used for the desired class in the "Decision Tree Learner" node. From the decision tree view depicted below we can see that most unmarried people have an income <=50K.
Coloring the nodes and leaves by for example "sex" (light blue for women and red for men), we can see (see figure below) that most "Wife" people are actually women and earn less than 50K per year.
The same color properties can be sent to the "Decision Tree Predictor" node via a "Color Appender" node, in order to observe the proportions of the different data groups of the test set in each node or leave of the tree.