gini-impurity — Cloudflare case

The team needs a reliable implementation for decision-tree training in a production-oriented ML workflow. A small implementation error can change the model input or metric enough to make a downstream result misleading.

Compute the Gini impurity of a binary label vector for a decision-tree node.

Compute $1-p^2-(1-p{)}^{2}1-p^2-(1-p)^2$ where $pp$ is the fraction of labels equal to 1. Labels must be 0 or 1; an empty node has impurity 0.0.

Implement solve(y).

Returns. Return a float between $00$ and $0.50.5$.

Examples

Example 1

Input:

solve([0, 0, 1, 1])

Output:

0.5

Example 2

Input:

solve([1, 1, 1, 1])

Output:

0.0

Example 3

Input:

solve([0, 1, 1, 1])

Output:

0.375