1 / 1

Compute Gini Impurity for a Split

Classic ML

Medium

Given the class labels sent to the left and right children of a decision-tree split, compute its weighted Gini impurity.

G(S)=1-\sum_{c=1}^{C}p_c^2

G_{\mathrm{split}}=\frac{N_L}{N}G(S_L)+\frac{N_R}{N}G(S_R)

Here, $p_c$ is the proportion of class $c$ in a node, $N_L$ and $N_R$ are the child sizes, and $N=N_L+N_R$ . An empty child has impurity zero. Return the weighted impurity as a Python float.

Loading visualization...

Examples

Input: y_left = [0, 0, 0], y_right = [1, 1, 1]

Output: 0.0

Explanation: Both child nodes contain only one class, so both impurities are zero.

Input: y_left = [0, 1], y_right = [0, 1]

Output: 0.5

Hint 1

Use np.unique(labels, return_counts=True) to obtain class frequencies.

Hint 2

Compute each node with 1.0 - np.sum(probabilities ** 2) before weighting.

Requirements

Compute class proportions independently for each child
Weight each child impurity by its share of all samples
Support binary and multiclass labels
Return a Python float

Constraints

The two label lists may have different lengths
A child list may be empty
Use NumPy only

Prev Next