=+47. The accompanying table shows the smallest value of SSResid for each number of predictors k (k 5 1, 2, 3, 4) for a regression problem in which y 5 cumulative heat of hardening in cement, x1 5 %

tricalcium aluminate, x2 5 % tricalcium silicate, x3 5 % aluminum ferrate, and x4 5 % dicalcium silicate:

Number of predictors k Predictors SSResid 1 x4 880.85 2 x1, x2 58.01 3 x1, x2, x3 49.20 4 x1, x2, x3, x4 47.86 In addition, n 5 13 and SSTo 5 2715.76.