Page 511 - Jolliffe I. Principal Component Analysis
P. 511
Index
476
supervised/unsupervised learning
248, 344, 345, 347–349, 372,
200
388, 390
SVD analysis, see maximum trend 148, 326, 336
covariance analysis removal of trend 76, 393
SVD see singular value tri-diagonal matrices 410
decomposition truncation of PC coefficients 67,
sweep-out components 403 293–296
switching of components 259 two-dimensional PC plots 2–4,
78–85, 130, 201–203, 212,
t-distribution/t-tests 186, 187, 191, 214–219, 234–236, 242–247,
193, 196, 197, 204, 205 258, 299
multivariate t-distribution 264, see also biplots, correspondence
364 analysis, interpretation
T-mode analysis 308, 398 of two-dimensional plots,
temperatures 22, 274, 316, 332 principal co-ordinate
air temperatures 71, 211, 302, analysis, projection pursuit
303, 329 two-stage PCA 209, 223
sea-surface temperatures 73,
211, 274, 275, 278–283, 286,
uncentred ‘covariances’ 290, 390
289, 310–314, 364, 396
uncentred PCA 41, 42, 349, 372,
tensor-based PCA 398
389, 391
three-mode factor analysis 397
units of measurement 22, 24, 65,
three-mode PCA 368, 397, 398 74, 211, 274, 374, 388, 391
time series 49, 56, 72, 74, 76, 128, upper triangular matrices, see
129, 148, 274, 290, 298–337,
lower triangular matrices
360, 365, 369, 370, 384, 393,
397, 398, 401
variable selection, see selection of
co-integration 330
variables
distributed lag model 337
variance inflation factors (VIFs),
moving averages 303, 368
see multicollinearities
seasonal dependence 300, 303,
variances for PCs, see PC
314, 315
variances
stationarity 300, 303, 304, 314,
316, 327, 330 variation between means 60, 85,
tests for randomness (white 96, 158
noise) 128 varimax rotation 153, 154,
see also autocorrelation, 162–165, 182, 188, 191, 238,
autoregressive processes, 270, 271, 274, 277–278
frequency domain PCs, red vector-valued data 129, 369, 370
noise, spectral analysis,
trend, white noise weighted PCA 21, 209, 241, 330,
T¨oplitz matrices 56, 303, 304 353, 382–385
transformed variables 64, 248, 374, weights
376, 377, 382, 386 exponentially decreasing 337,
logarithmic transformation 24, 368, 384

