How to read the decision tree

tonyboy9

In Studio, under File, New process, Learn, 2. Modeling.
Using the Titanic training data set, the decision tree is 
built, with the Explanation: 

From our resulting decision-tree, we can clearly see that survival was not coincidental at all. In fact, it tells us that as a female passenger, with a small family, you were really lucky - at least if you were holding an expensive ticket!
In the screen shot, how do you read the decision tree to tell you all that?

Telcontar120

It is a terminal node, which contains a prediction (in this case "Yes") and also a numerical confidence for that prediction based on the count of records in each class at that node.  If you look at the DT model output in RapidMiner in the list view instead of the graphical view, this is a bit clearer.

sara20

@tonyboy9

Hello

The root is "No of parents and child on board" so there is 2 parts for less than 4500 or more than 4500 then if the answer is less than 4500 you will find "No of sibling and ..." so now you know the root and other parts are leaves.

I hope this helps

tonyboy9

Thank you for responding, Sara. Perhaps you can tell what does the 'yes' box with the blue mean?

Telcontar120

It is a terminal node, which contains a prediction (in this case "Yes") and also a numerical confidence for that prediction based on the count of records in each class at that node.  If you look at the DT model output in RapidMiner in the list view instead of the graphical view, this is a bit clearer.