Geometric Mean Ratios

Problem

Aggregate forecast accuracy multiplicatively. Compute the ratio forecast/actual for every point with a nonzero actual, then return the geometric mean of those ratios: the product raised to the power 1/k.

Return 0.0 when there are no usable ratios.

Hint

A value of 1.0 means the forecast is right on average in relative terms.

Test cases

case 1

in: [[100,200],[110,190]]

expected: 1.0222524150130436

case 2

in: [[50],[55]]

expected: 1.1

case 3

in: [[10,10],[10,10]]

expected: 1

case 4

in: [[4,8],[5,6]]

expected: 0.9682458365518543

Starter codePython

def geometric_mean_ratios(actual, forecast):
    # Your code here
    pass

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