Transformed Outcome

Problem

Compute the transformed outcome for a constant propensity p: y * (t - p) / (p * (1 - p)), clipping p to [1e-6, 1 - 1e-6].

Empty input returns [].

Hint

The transformed outcome is unbiased for the treatment effect under the true propensity.

case 1

in: [[1,1,0],[1,1,0],[0.5,0.5,0.5]]

expected: [2,2,0]

case 2

in: [[1,0],[1,0],[0.8,0.8]]

expected: [1.25,0]

case 3

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

expected: [1.000001000001,1.000001000001]

case 4

in: [[],[],[]]

expected: []

Starter codePython

def transformed_outcome(y, treatment, propensity):
    # Your code here
    pass

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