Significance vs effect size

For 3 key comparisons, report a p-value AND an effect size, and decide which matters. Concepts: p-values, effect size (difference in means/rates, Cohen's d), multiple comparisons Project A: Mann-Whitney + Cohen's d per numeric feature. Lesson from your run: monthly_payment_proxy has p = 0.003 but Cohen's d = 0.04 (negligible) → 'significant' is not 'useful'. Keep rule = p < 0.05 AND |d| ≥ 0.2. Project B: Weekday vs weekend mean load difference (MW and %).