Decision trees' "top node"


I’m not sure why the top node is regarded as where splits haven’t happened. For example, in the below tree, the top node seems to be the result of a previous split that result in the "“214,106” grouping.

top node

In short, if the data described in the top node has been splitted, why is the node considered “top”? If it’s not been splitted yet, where does such grouping as “214,106” come from?

Thanks a lot.:slight_smile: