MediumCalculusca-352

Weighted MAE Gradient

Problem

For a weighted mean absolute error, the subgradient for each pair is w_i * sign(pred_i - target_i) with sign 0 at exact ties.

Given predictions, targets, and per-example weights, return the subgradient list.

Hint

Compute the sign first, then scale by the example weight.

case 1

in: [[1,2],[2,2],[1,1]]

expected: [-1,0]

case 2

in: [[0,5],[0,0],[2,3]]

expected: [0,3]

case 3

in: [[1,-1],[-1,1],[0.5,0.5]]

expected: [0.5,-0.5]

Starter codePython

def weighted_mae_gradient(preds, targets, weights):
    # Your code here
    pass

3 test cases · run your code against them in the editor