The root node of a decision tree is built via calculating the gini impurity of each root tree for each feature. The the feature that generates the lowest gini impurity is then chosen as the root.

flowchart TD A[Loves popcorn\nGini impurity: 0.32] A --> |True| B[Loves the song] A --> |False| C[Deos not love the song]
flowchart TD A[Loves pop\nGini impurity: 0.1] A --> |True| B[Loves the song] A --> |False| C[Deos not love the song]
flowchart TD A[Loves cats\nGini impurity: 0.48] A --> |True| B[Does not loves the song] A --> |False| C[Loves the song]

References