Paired t Confidence Interval on CV Folds

Problem

Return the paired confidence interval for the mean difference between model A and model B across cross-validation folds: mean(d) +/- t_crit * sd(d)/sqrt(k) with d = fold_a - fold_b. Return [0.0, 0.0] when fewer than two folds or lengths differ, and [mean(d), mean(d)] when the differences are constant.

Hint

Treat the folds as paired observations and use the supplied t critical value.

Test cases

case 1

in: [[0.9,0.8,0.85,0.87],[0.8,0.75,0.8,0.79],2.353]

expected: [0.0411817531761559,0.09881824682384406]

case 2

in: [[1,2,3],[1,2,3],2]

expected: [0,0]

case 3

in: [[0.5,0.6],[0.4,0.5],12.706]

expected: [0.09999999999999998,0.09999999999999998]

case 4

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

expected: [0,0]

case 5

in: [[],[],2]

expected: [0,0]

Starter codePython

def paired_ci_cv(fold_a, fold_b, t_crit):
    # Your code here
    pass

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