Paired t-test on cross-validation folds

Test whether two models differ on the same folds

Given per-fold scores of two models evaluated on the same folds, the paired t-test looks at the per-fold differences d_i = a_i - b_i. The statistic is

t = mean(d) / (sd(d) / sqrt(n)), with sd using n - 1 in the denominator.

Implement paired_t_statistic(a, b) and return t as a float.

Constraints

• a and b are 1-D arrays of the same length n >= 2. Otherwise raise ValueError.
• If the differences have zero spread, the statistic is undefined, so raise ValueError.
• Do not modify the inputs.

Hints