R语言主成分分析

我的数据是果实在生长过程中测量性状,共测量8个性状,不同时间点测量共计8次,每次300个样品,请问这个如何做主成分分析。如果按照你的方法,做出来只是8个点。attachments-2018-06-VJIPFK8S5b241119b81a7.jpeg

> cz<-read.table("x:/Yaping.Ma/cz.txt",header=TRUE,sep="\t")
> head(cz)                  
     x1    x2   x3   x4   x5    x6   x7 x8
1 30.06 11.66 2.02 9.68 4.06 21.41 3.55  0
2 32.18 13.42 2.29 7.96 5.42 22.85 2.59  0
3 28.31 11.56 1.95 7.78 4.07 16.36 3.42  0
4 28.63 11.98 1.99 7.42 4.60 16.01 2.79  0
5 27.42 11.58 1.83 7.30 3.26 18.97 5.07  0
6 24.75 10.25 1.24 6.21 3.10 17.70 4.06  0
> pca1<-princomp(cz,cor=T)     
> summary(pca1,loadings = T)   
Importance of components:

                          Comp.1    Comp.2    Comp.3     Comp.4
Standard deviation     2.1585794 1.2076351 0.9174367 0.75943666
Proportion of Variance 0.5824331 0.1822978 0.1052113 0.07209301
Cumulative Proportion  0.5824331 0.7647309 0.8699422 0.94203521
                           Comp.5     Comp.6     Comp.7
Standard deviation     0.52737553 0.36470473 0.22931097
Proportion of Variance 0.03476562 0.01662619 0.00657294
Cumulative Proportion  0.97680083 0.99342702 0.99999996
                             Comp.8
Standard deviation     5.305783e-04
Proportion of Variance 3.518916e-08
Cumulative Proportion  1.000000e+00

 

Loadings:
   Comp.1 Comp.2 Comp.3 Comp.4 Comp.5 Comp.6 Comp.7 Comp.8
x1 -0.424 -0.187               -0.247  0.819 -0.227       
x2 -0.442 -0.169 -0.103 -0.167        -0.349 -0.323 -0.714
x3 -0.432 -0.233        -0.139                0.854       
x4         0.496 -0.859                                   
x5 -0.427 -0.258 -0.164               -0.347 -0.323  0.693
x6 -0.316  0.279  0.150  0.791 -0.366 -0.202              
x7 -0.190  0.573  0.395 -0.554 -0.388                0.101
x8 -0.340  0.404  0.212         0.796  0.178             
> library(scatterplot3d)  
> PCA1=pca1$loadings[,1]    
> PCA2=pca1$loadings[,2]
> PCA3=pca1$loadings[,3]
> colors=rainbow(24)       
> s3d=scatterplot3d(PCA1,PCA2,PCA3
+                   #, highlight.3d = TRUE
+                   , col.axis = "blue",angle = 40,
+                   color=colors[seq(1,24,3)], main = "Principal component analysis", pch = ' ')
> s3d$points(PCA1,PCA2, PCA3, pch =15:22,
+            cex = 2,col=colors[seq(1,24,3)])
> legend(s3d$xyz.convert(0.2, 0.2, -0.4), pch = 15:22, yjust=0,
+        legend =colnames(cz), cex = .7,col=colors[seq(1,24,3)],bty="n")
请先 登录 后评论

1 个回答

祝让飞 - 生物信息工程师

你这是相当于对8个性状做主成分分析,如果要对300个样本的话,你的数据矩阵应该转置 

请先 登录 后评论
  • 1 关注
  • 0 收藏,2678 浏览
  • 马亚平 提出于 2018-06-16 03:20