| reconst {ade4} | R Documentation |
Generic Function for the reconstitution of data from a principal component analysis or a correspondence analysis
reconst (dudi, ...) ## S3 method for class 'pca': reconst(dudi, nf = 1, ...) ## S3 method for class 'coa': reconst(dudi, nf = 1, ...)
dudi |
an object of class dudi used to select a method: pca or coa |
nf |
an integer indicating the number of kept axes for the reconstitution |
... |
further arguments passed to or from other methods |
returns a data frame containing the reconstituted data
Daniel Chessel
Anne B Dufour dufour@biomserv.univ-lyon1.fr
Gabriel, K.R. (1978) Least-squares approximation of matrices by additive and multiplicative models. Journal of the Royal Statistical Society, B , 40, 186–196.
data(rhone)
dd1 <- dudi.pca(rhone$tab, nf = 2, scann = FALSE)
rh1 <- reconst(dd1, 1)
rh2 <- reconst(dd1, 2)
par(mfrow = c(4,4))
par(mar = c(2.6,2.6,1.1,1.1))
for (i in 1:15) {
plot(rhone$date, rhone$tab[,i])
lines(rhone$date, rh1[,i], lty = 2)
lines(rhone$date, rh2[,i], lty = 1)
scatterutil.sub(names(rhone$tab)[i], 2, "topright")}
data(chats)
chatsw <- data.frame(t(chats))
chatscoa <- dudi.coa(chatsw, scann = FALSE)
model0 <- reconst(chatscoa, 0)
round(model0,3)
round(chisq.test(chatsw)$expected,3)
chisq.test(chatsw)$statistic
sum(((chatsw-model0)^2)/model0)
effectif <- sum(chatsw)
sum(chatscoa$eig)*effectif
model1 <- reconst(chatscoa, 1)
round(model1, 3)
sum(((chatsw-model1)^2)/model0)
sum(chatscoa$eig[-1])*effectif
> library(ade4)
> ### Name: reconst
> ### Title: Reconstitution of Data from a Duality Diagram
> ### Aliases: reconst reconst.pca reconst.coa
> ### Keywords: multivariate
>
> ### ** Examples
>
> data(rhone)
> dd1 <- dudi.pca(rhone$tab, nf = 2, scann = FALSE)
> rh1 <- reconst(dd1, 1)
> rh2 <- reconst(dd1, 2)
> par(mfrow = c(4,4))
> par(mar = c(2.6,2.6,1.1,1.1))
> for (i in 1:15) {
+ plot(rhone$date, rhone$tab[,i])
+ lines(rhone$date, rh1[,i], lty = 2)
+ lines(rhone$date, rh2[,i], lty = 1)
+ scatterutil.sub(names(rhone$tab)[i], 2, "topright")}
>
> data(chats)
> chatsw <- data.frame(t(chats))
> chatscoa <- dudi.coa(chatsw, scann = FALSE)
> model0 <- reconst(chatscoa, 0)
> round(model0,3)
age1 age2 age3 age4 age5 age6 age7 age8
f0 6.285 6.638 4.379 2.331 1.624 1.059 1.130 1.554
f12 11.816 12.480 8.232 4.381 3.054 1.992 2.124 2.921
f34 30.924 32.661 21.542 11.466 7.992 5.212 5.559 7.644
f56 14.331 15.136 9.983 5.314 3.703 2.415 2.576 3.542
f78 13.828 14.605 9.633 5.127 3.573 2.331 2.486 3.418
f9a 8.799 9.294 6.130 3.263 2.274 1.483 1.582 2.175
fbc 1.760 1.859 1.226 0.653 0.455 0.297 0.316 0.435
fcd 1.257 1.328 0.876 0.466 0.325 0.212 0.226 0.311
> round(chisq.test(chatsw)$expected,3)
age1 age2 age3 age4 age5 age6 age7 age8
f0 6.285 6.638 4.379 2.331 1.624 1.059 1.130 1.554
f12 11.816 12.480 8.232 4.381 3.054 1.992 2.124 2.921
f34 30.924 32.661 21.542 11.466 7.992 5.212 5.559 7.644
f56 14.331 15.136 9.983 5.314 3.703 2.415 2.576 3.542
f78 13.828 14.605 9.633 5.127 3.573 2.331 2.486 3.418
f9a 8.799 9.294 6.130 3.263 2.274 1.483 1.582 2.175
fbc 1.760 1.859 1.226 0.653 0.455 0.297 0.316 0.435
fcd 1.257 1.328 0.876 0.466 0.325 0.212 0.226 0.311
> chisq.test(chatsw)$statistic
X-squared
74.53507
> sum(((chatsw-model0)^2)/model0)
[1] 74.53507
> effectif <- sum(chatsw)
> sum(chatscoa$eig)*effectif
[1] 74.53507
> model1 <- reconst(chatscoa, 1)
> round(model1, 3)
age1 age2 age3 age4 age5 age6 age7 age8
f0 7.450 7.167 4.164 2.102 1.381 0.892 0.658 1.187
f12 14.896 13.877 7.665 3.777 2.411 1.548 0.876 1.951
f34 40.924 37.196 19.701 9.505 5.905 3.771 1.505 4.493
f56 15.760 15.784 9.720 5.033 3.405 2.209 1.997 3.092
f78 6.961 11.490 10.897 6.474 5.006 3.320 5.270 5.582
f9a 2.776 6.562 7.239 4.444 3.531 2.351 4.024 4.073
fbc 0.063 1.089 1.538 0.985 0.809 0.541 1.004 0.970
fcd 0.171 0.835 1.076 0.679 0.551 0.368 0.666 0.653
> sum(((chatsw-model1)^2)/model0)
[1] 31.68349
> sum(chatscoa$eig[-1])*effectif
[1] 31.68349
>
>
>
>