R语言实现PCA降维

R代码实现PCA降维过程

链接: http://blog.sciencenet.cn/blog-3448646-1270918.html
数据集是经典的鸢尾花
做简单的记录，只附代码和结果，
下面展示一些 内联代码片。

具体自行操作，可交流

install.packages("ggplot2")
library(ggplot2)
df <- iris[c(1, 2, 3, 4)]
head(df)
df_pca <- prcomp(df) #计算主成分
df_pcs <-data.frame(df_pca$x,Species=iris$Species)  
head(df_pcs,3)  #查看主成分结果
#3.2 ggplot2 绘制PCA图
#1) Species分颜色
ggplot(df_pcs,aes(x=PC1,y=PC2,color=Species))+ geom_point()
#2）去掉背景及网格线
ggplot(df_pcs,aes(x=PC1,y=PC2,color=Species))+ 
  
  geom_point()+ 
  
  theme_bw() +
  
  theme(panel.border=element_blank(),
        
        panel.grid.major=element_blank(),
        
        panel.grid.minor=element_blank(),
        
        axis.line= element_line(colour = "black"))
#3) 添加PC1  PC2的百分比
percentage<-round(df_pca$sdev/sum(df_pca$sdev) * 100,2)

percentage<-paste(colnames(df_pcs),"(",
                  
                  paste(as.character(percentage), "%", ")", sep=""))

ggplot(df_pcs,aes(x=PC1,y=PC2,color=Species))+
  
  geom_point()+ 
  
  xlab(percentage[1]) +
  
  ylab(percentage[2])



#4) 添加置信椭圆
ggplot(df_pcs,aes(x=PC1,y=PC2,color = Species))+ 
  
  geom_point()+stat_ellipse(level = 0.95, show.legend = F) + 
  
  annotate('text', label = 'setosa', x = -2, y = -1.25, 
           
           size = 5, colour = '#f8766d') +
  
  annotate('text', label = 'versicolor', x = 0, y = - 0.5, 
           
           size = 5, colour = '#00ba38') +
  
  annotate('text', label = 'virginica', x = 3, y = 0.5,
           
           size = 5, colour = '#619cff')
#5) 查看各变量对于PCA的贡献

df_r <- as.data.frame(df_pca$rotation)

df_r$feature <- row.names(df_r)

df_r 
#贡献度绘图

ggplot(df_r,aes(x=PC1,y=PC2,label=feature,color=feature )) + 
  
  geom_point()+ geom_text(size=3)
#四 PCA绘图汇总展示

ggplot(df_pcs,aes(x=PC1,y=PC2,color=Species )) +
  
  geom_point()+xlab(percentage[1]) +
  
  ylab(percentage[2]) + 
  
  stat_ellipse(level = 0.95, show.legend = F) +
  
  annotate('text', label = 'setosa', x = -2, y = -1.25,
           
           size = 5, colour = '#f8766d') +
  
  annotate('text', label = 'versicolor', x = 0, y = - 0.5, 
           
           size = 5, colour = '#00ba38') +
  
  annotate('text', label = 'virginica', x = 3, y = 0.5, 
           
           size = 5, colour = '#619cff') +
  
  labs(title="Iris PCA Clustering", 
       
       subtitle=" PC1 and PC2 principal components ",    
       
       caption="Source: Iris") + theme_classic()


screeplot(df_pca,type = "line",lwd=3)#碎石图


!