Decision Tree Stump

Implement a decision tree stump (depth-1 tree) for binary classification. Given a dataset with one numeric feature and binary labels (0 or 1), find the best threshold to split on that minimizes Gini impurity.

Return a function that classifies a value as 0 or 1, along with the threshold.

Example

features = [1, 2, 3, 4, 5]
labels = [0, 0, 1, 1, 1]
threshold, left_label, right_label = decision_stump(features, labels)
# threshold=2.5, left_label=0, right_label=1