Principal component analysis biplots - Understanding Biplots - page 96

Ta b l e 3 . 9 The Ocotea data set.

Sample Species VesD VesL FibL RayH RayW NumVes

1 Obul 79 383 941 333 30 17

2 Obul 78 346 961 223 24 31

3 Obul 82 361 1039 316 27 25

4 Obul 79 324 1048 369 29 26

5 Obul 85 418 1051 347 34 14

6 Obul 111 448 1096 379 40 13

7 Obul 76 320 1130 347 29 13

8 Obul 103 371 1165 326 26 10

9 Obul 129 406 1165 428 44 11

10 Obul 74 281 1175 324 26 11

11 Obul 102 567 1221 395 40 11

12 Obul 95 415 1225 416 38 10

13 Obul 91 372 1234 375 26 11

14 Obul 113 314 1253 466 23 10

15 Obul 93 541 1267 347 34 14

16 Obul 94 437 1271 336 36 10

17 Obul 119 359 1280 412 32 11

18 Obul 104 387 1290 381 22 12

19 Obul 114 569 1369 568 52 11

20 Obul 141 621 1527 419 34 15

21 Oken 147 402 1391 440 32 9

22 Oken 142 393 1468 443 35 6

23 Oken 125 322 1530 459 34 11

24 Oken 156 401 1588 512 42 11

25 Oken 162 502 1591 369 42 8

26 Oken 103 378 1655 441 34 11

27 Oken 126 414 1759 459 42 8

28 Opor 122 346 981 393 40 14

29 Opor 139 133 993 342 33 14

30 Opor 130 368 1005 356 39 16

31 Opor 127 331 1027 473 38 20

32 Opor 112 309 1044 358 47 8

33 Opor 115 352 1048 300 36 14

34 Opor 130 471 1072 409 39 15

35 Opor 153 419 1077 392 48 20

36 Opor 135 370 1104 531 38 15

37 Opor 130 325 1166 428 36 12

5.23 to 214.05, the data were pre-centred and pre-scaled to unit variances. The biplot
axes are calibrated in terms of the original measurement units. The usefulness of an
initial PCA biplot could be improved by augmenting it with the interpolated centroids of
the three Ocotea species. The mean vectors of these three species are given in Table 3.10.
To obtain the biplot in Figure 3.23 we first represent Table 3.10 as an R object
newsamples in the form of a 3 × 6 matrix. The following function call creates the
biplot in Figure 3.23: