PCR(Principal Component Regression)とPLSR(Partial Least Squares Regression

• 
• 重回帰 Multiple Linear Regression
  • 

# 重回帰 multiple linear regression
n<-6
m<-4
X<-matrix(runif(n*m),n,m)
m2<-2
Y<-matrix(runif(n*m2),n,m2)
B<-solve(t(X)%*%X)%*%t(X)%*%Y
Y-X%*%B
# n<m とすると解けない
m<-n+1
X<-matrix(runif(n*m),n,m)
m2<-2
Y<-matrix(runif(n*m2),n,m2)
B<-solve(t(X)%*%X)%*%t(X)%*%Y
Y-X%*%B

• PCR
  • こちらに書き直し