Huber IRLS Weight

Problem

Compute the Huber IRLS weight for a residual.

Return 1.0 when abs(residual) <= delta, otherwise delta / abs(residual). Outliers receive a weight that decays like the reciprocal of their residual.

Hint

Inliers are weighted fully; outliers are downweighted proportionally.

case 1

in: [0.5,1]

expected: 1

case 2

in: [2,1]

expected: 0.5

case 3

in: [-3,2]

expected: 0.6666666666666666

case 4

in: [0,1]

expected: 1

Starter codePython

def huber_irls_weight(residual, delta):
    # Your code here
    pass

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