Compare the two best models that you have found in Problem 10.24 in terms of \(C_{p}\) by calculating the confidence intervals on the mean of the response for all points in the original data set. Based on a comparison of these confidence intervals, which model would you prefer? Now calculate the PRESS statistic for these models. Which model would PRESS indicate is likely to be the best for predicting new response observations?

Problem 10.24

Table B. 14 presents data on the transient points of an electronic inverter. Use all possible regressions and the CpCp${\mathrm{????}}_{\mathrm{????}}$ criterion to find an appropriate regression model for these data. Investigate model adequacy using residual plots.

Observation Number 11 1234567RROUNDIN 3 8 3 4 8 10 8 8 6 4 10 10 16 3 12 10 8 3 3 4 16 5 17 2 18 10 15 19 15 20 15 21 10 22 3 23 6 24 25 3 2 38625534400 en en en 2 m en 3332468 1 mm 8422322326 53064736882256648633 X3 Xs 3864655 000 0.787 0.293 1.710 12 0 0.203 0 0.806 0 4.713 3 25 4 25 12 8 25 50 50 50 50 33333388888 0.607 9.107 9.210 25 1.365 25 4.554 25 0.293 50 2.252 9.167 0.694 0.379 0.485 50 3.345 50 0.208 75 0.201 75 0.329 75 4.966 75 1.362 75 1.515 75 0.751