Whenever you're ready, walk me through how you'd approach this.
How would you determine whether a change in model performance is statistically significant? Treat the two model versions as paired measurements evaluated on the same ten folds, and use the performance scores in the given data. State the hypotheses, calculate the paired test statistic and two-sided p-value at the 5% level, quantify uncertainty with a 95% confidence interval, and explain the practical interpretation. Assume the fold-level differences are approximately independent and normally distributed.