Use the SYMPTOM_TEXT Variable to Model

To use the SYMPTOM_TEXT variable to create a model, complete the following steps:

Right-click the Text Miner — CST node, and select Copy from the menu. Right-click an empty space in the diagram workspace and select Paste. Connect the Text Miner — COSTART node to the Text Miner — CST node.

This second Text Miner — CST node will be used to analyze the SYMPTOM_TEXT variable. SYMPTOM_TEXT will be the default parse variable because it is the longest text field in the data set. You need to specify COSTRING as a parse variable as well.
Select the second Text Miner — CST node. Click the button for the Variables property in the Properties panel.
In the Variables window, set the following:
- Set the Use value of SYMPTOM_TEXT to Yes.
- Set the Use value of costring to Yes.
- Set the Use value of serious to Yes.
Click OK.
Set the following properties in the Properties Panel:
- Set Compute SVD to Yes.
- Set SVD Resolution to Low.
- Set Term Weight to Mutual Information.
Right-click the new Text Miner node, and select Run. Click Yes in the Confirmation dialog box. Click OK in the Run Status dialog box.
From the Model tab, drag and drop a Decision Tree node into the diagram workspace. Connect the new Text Miner node to the Decision Tree node. You will use the decision tree to see whether text mining the original text can do a better job of predicting serious events than just mining the COSTART terms.
Right-click the new Decision Tree node and select Rename. Type Decision Tree — ST, where ST stands for Symptom Text, in the Node Name text box. Click OK.
Click the button for the Variables property in the Decision Tree — ST properties panel. The Variable window opens.
Click and scroll to select all of the _ROLL_ variables, and then set the _ROLL_ Use values to No.
Click OK to save your changes.
Right-click the Decision Tree — ST node, and select Run. Click Yes in the Confirmation dialog box. Click OK in the Run Status dialog box when the node has finished running.
From the Model tab, drag and drop a Decision Tree node and connect it to the Text Miner — CST node.
Right-click the new Decision Tree node, and select Rename. Type Decision Tree — CST, where CST stands for COSTART and Symptom Text, in the Node Name box. Click OK. This node will let you see how well you can predict serious events with all the information available to you. Use the default settings for the node.