0

I want to ask about the eigen function of R. I am currently doing a project of NBA team analysis. I am trying to figure out correlation effect of two players lineup and who effect positively in offense and who effect negatively in offense. I am trying to reduce dimensions starting from using eigen() function. First off, I made a vector size of 100 and transformed it as a matrix size of 10 by 10.

a=c(107.4,105.0,106.0,105.4,106.5,107.1,103.8,105.1,105.8,108.3,
    105.0,106.3,102.5,101.3,104.9,105.6,100.8,105.8,103.2,113.0,
    106.0,102.5,105.0,100.1,102.6,107.9,101.3,101.9,102.1,105.8,
    105.4,101.3,100.1,104.9,104.2,105.8,101.8,103.5,102.4,102.9,
    106.5,104.9,102.6,104.2,106.0,106.6,98.3,101.1,105.6, 103.9,
    107.1,105.6,107.9,105.8,106.6,107.9,104.6,103.8,106.2,101.5,
    103.8,100.8,101.3,101.8,98.3 ,104.6,103.4,100.8,NA   ,103.2,
    105.1,105.8,101.9,103.5,101.1,103.8,100.8,103.5,103.7,98.5,
    105.8,103.2,102.1,102.4,105.6,106.2,NA   ,103.7,105.1,99.5,
    108.3,113.0,105.8,102.9,103.9,101.5,103.2,98.5, 99.5, 107.5)
b=matrix(a,nrow=10)

The order is "kuzma","ingram","hart","ball","pope","lebron","mcgee","stephenson","chandler","rondo"

These are player names of the LA Lakers and the [1,1] component means offensive rating of Kuzma. [1,5] means offensive rating of Kuzma and Pope.

b[7,9]=mean(b[7,],na.rm=T)

7th player mcgee and 9th player chandler didn't play each other for a long time so I set it the mean offensive rating of mcgee and other players.

b[9,7]=b[7,9]

The offensive rating of Mcgee and Chander is the same as the one of Chandler and Mcgee.

corrmat=cor(b)

dimnames(corrmat)=list(c("kuzma","ingram","hart","ball","pope","lebron","mcgee","stephenson","chandler","rondo"))

eigen(corrmat)

The eigen function's result is way different from what I expected. The eigenvalue goes down from 3.2, 2.8, 1.3 and so on. Is this meaning the component Kuzma explains more than Ingram and Ingram more than Hart and so on? Lebron James, the most famous player and who actually has high offensive rating has low eigenvalue. I want the way of explaining the result of this eigen function in R.

Nick Cox
  • 48,377
  • 8
  • 110
  • 156
Damelim
  • 1
  • 1
    I don't see that you're really asking about the `eigen()` function in R. Your question seems to be why you get results from PCA that puzzle you. But there seems to be a statistical misunderstanding here. Kuzma, Ingram, etc. are variables in your data as I understand it, but they aren't also names of components. – Nick Cox Feb 21 '19 at 10:32
  • Consider starting your exploration of this topic by reading the answers at https://stats.stackexchange.com/questions/2691/making-sense-of-principal-component-analysis-eigenvectors-eigenvalues. – whuber Feb 21 '19 at 13:05

0 Answers0