Model A has the lower total SSE across x = 20 to 80, but its largest residuals occur from x = 70 to 80. Model B has slightly higher total SSE but much smaller residuals from x = 70 to 80. Next month all predictions will be in that upper interval. Which model should be used?