Statistics – predict from a tree model

 In Statistics & methods

This is a blog tip from one of our SPSS Statistics user, he really recommend to use the score wizard. I have in this example made a tree model and then you will see how simple it is to save and then use the score model for future data.

So here is an example when I do a tree analysis, to understand the most important combinations of characteristics that explain who is most likely answering an offer campaign (YES compared to NO in the variable “response” below).

The tree analysis:

The tree result:

In the dialogbox of the tree analysis you can save scores (like you do in regression) if you click on the “Save button”, and this models definition of the scores rules can be saved into a XML-file (see below):

Now to the prediction.

We want to use this saved XML-file upon a quite new data set to predict how likely it is that these 10 people answer YES on the offer campaign.

Choose the command:

Then choose the saved XML-file and choose the target category (YES in my example, that is coded as 2):

Here is the prediction in the new data file:

So the first person is most likely to answer YES to our Campaign, the chance is 42%.

Greetings Gunilla Rudander


error: Content is protected !!