pca code.txt

##############################
# PCA example                #
##############################

data2=read.table("C:\\Users\\yxiong\\Desktop\\data2.txt",header=T)
attach(data2)
# standardize the data (pca isnot scale invariant)
  data2.scaled=apply(data2,2,scale)
  correlation=round(cor(data2.scaled),2)
  correlation

# apply PCA
  pca.res=prcomp(data2.scaled,retx=T)
  pca.res
  summary(pca.res)
  plot(pca.res) #display the variances
  #windows()
  #screeplot(pca.res,type="lines")

pca.res$x[1:5,1:2] #total 34 rows, display pc1 and pc2 here
plot(pca.res$x[,1:2], pch="")
text(pca.res$x[,1:2])
abline(v=0, lty=2)
abline(h=0, lty=2)

  corr1 = cor(pca.res$x) #check the corr of principal component
  round(corr2, 1)
  biplot(pca.res)
  abline(0, pca.res$rotation[2,2]/pca.res$rotation[2,1])

##############################
# factor analysis example    #
##############################

 res2 = factanal(data2, factors=2, rotation="varimax") #orthogonal rotation , for oblique using promax
 res2                                  
#May be unrealistic to assume that factors are uncorrelated. One may obtain a better fit by dropping this assumption.