4. CASE Switch

<p><strong>CASE Switch</strong></p><p>This workflow implements a CASE Switch to control workflow execution. The goal is to classify the car makers based on the cars they produce. The target class is the "make" column. The choice of which prediction algorithm is used is left to the end user and can be selected through the <em>Value Selection Widget</em> node. The choices are </p><ul><li><p> a set of rules (top branch), </p></li><li><p>a Decision Tree (middle branch) or, </p></li><li><p>a PNN (bottom branch).</p></li></ul><p>Depending on which predictive model has been chosen, the corresponding port of the <em>CASE Switch Start</em> node is activated and the respective model is trained and applied. The predictions are collected in the <em>CASE Switch End </em>node (data ports), and performances on the training set are evaluated at the end with a <em>Scorer </em>node. The important part is to produce predictions in an output column that is named the same on all three branches. The selected trained model is collected with a separate <em>CASE Switch End</em> node (model ports) and written to a file.</p>