Lecture 25
Looking at the plots of the predictors against the raw residuals
tells if there is curvature but need not suggest the appropriate
transformation. Looking at the partial residuals should suggest the
appropriate transformation. We will look at the clear-cut example
from Mendenhall and Sincich: A Second Course in Statistics,
Regression Analysis, which models the price of coffee on the demand for
coffee. Then we will look at the pollution data from Rice again.
demand price advertisement
1190 3.0 1
1033 3.2 1
897 3.4 1
789 3.6 1
706 3.8 1
595 4.0 1
512 4.2 1
433 4.4 1
395 4.6 1
304 4.8 1
243 5.0 1
1124 3.0 0
974 3.2 0
830 3.4 0
702 3.6 0
619 3.8 0
529 4.0 0
451 4.2 0
359 4.4 0
296 4.6 0
247 4.8 0
194 5.0 0
Day WindSped Temp Humidity Insolat Oxidant
1 50 77 67 78 15
2 47 80 66 77 20
3 57 75 77 73 13
4 38 72 73 69 21
5 52 71 75 78 12
6 57 74 75 80 12
7 53 78 64 75 12
8 62 82 59 78 11
9 52 82 60 75 12
10 42 82 62 58 20
11 47 82 59 76 11
12 40 80 66 76 17
13 42 81 68 71 20
14 40 85 62 74 23
15 48 82 70 73 17
16 50 79 66 72 16
17 55 72 63 69 10
18 52 72 61 57 11
19 48 76 60 74 11
20 52 77 59 72 9
21 52 73 58 67 5
22 48 68 63 30 5
23 65 67 65 23 4
24 53 71 53 72 7
25 36 75 54 78 18
26 45 81 44 81 17
27 43 84 46 78 23
28 42 83 43 78 23
29 35 87 44 77 24
30 43 92 35 79 25