Page 509 - Jolliffe I. Principal Component Analysis
P. 509
Index
474
289, 290, 291
see also distance/dissimilarity
SCoTLASS (simplified component
measures
technique - LASSO) simple components 280–287, 291
280–283, 287–291 simplicity/simplification 269–271,
scree graph 115–118, 125, 126, 274, 277–286, 403, 405
130–132, 134, 135 simplified PC coefficients 66, 67,
selection of subsets of PCs 76, 77
in discriminant analysis 201, see also approximations to PCs,
202, 204–206 discrete PC coefficients,
in latent root regression 180, 181 rounded PC coefficients
in PC regression 168, 170–177, simultaneous components 361
196–198, 202, 205, 245 singular spectrum analysis (SSA)
see also how many PCs, rules 302–308, 310, 316
for selecting PCs singular value decomposition
selection of variables (SVD) 7, 29, 44–46, 52, 59,
in non-regression contexts 13, 101, 104, 108, 113, 120, 121,
27, 38, 111, 137–149, 186, 129, 172, 173, 226, 229, 230,
188, 191, 198, 220, 221, 260, 253, 260, 266, 273, 353, 365,
270, 286, 288, 290, 293–295, 366, 382, 383
376 comparison of SVDs 362
stepwise selection/backward computation based on SVD 46,
elimination algorithms 142, 173, 412, 413
144, 145, 147 generalized SVD 46, 342, 383,
see also principal variables, 385, 386
regression analysis (variable multitaper frequency domain
selection) SVD (MTM-SVD) 302, 311,
self-consistency 20, 378, 379 314, 316
sensible PCA 60 size and shape PCs 53, 57, 64, 67,
sensitivity matrix 240 68, 81, 104, 297, 298, 338,
sensitivity of PCs 232, 252, 343–346, 355, 356, 388, 393,
259–263, 278 401
shape and size PCs, see size and see also contrasts between
shape PCs variables, interpretation
Shapiro-Wilk test 402 of PCs, patterned
shrinkage methods 167, 178–181, correlation/covariance
264, 288 matrices
signal detection 130, 304, 332 skewness 219, 372
signal processing 303, 317, 395 smoothing and interpolation 274,
signal to noise ratio 337, 388, 401 316, 318, 320, 322, 324–326,
SIMCA 207–208, 239 334, 335, 377–379
similarity measures of spatial data 334, 335, 364, 365
between configurations 38 lo(w)ess 326
between observations 79, 89, splines 320, 322, 331, 377, 378,
106, 210-212, 339, 390 387
between variables 89, 213, 391 sparse data 331

