11.8 Correlation Results Following Additive Logratio or Centred

# Fig.11.6.: Plot correlation matrix with log-ratio transformation library(StatDA) data(chorizon) x=chorizon[,c("Al2O3","CaO","Fe2O3","K2O","MgO","MnO","Na2O","P2O5","SiO2","TiO2")] pdf("fig-11-6.pdf",width=7,height=7) par(mfrow=c(1,1),mar=c(4,4,2,0)) x1=x/x[,"TiO2"] x1.obj=log10(x1[,1:9]) CorCompare(cor(x1.obj),cor(log10(x[,1:9])), labels1=dimnames(x)[[2]],labels2=dimnames(x)[[2]], method1="additive logratio transformation",method2="log-transformation", ndigits=2) dev.off()

# Fig.11.7.: Plot correlation matrix with log-centring transformation library(StatDA) data(chorizon) x=chorizon[,c("Al2O3","CaO","Fe2O3","K2O","MgO","MnO","Na2O","P2O5","SiO2","TiO2","LOI")] pdf("fig-11-7.pdf",width=7,height=7) par(mfrow=c(1,1),mar=c(4,4,2,0)) xgeom=10^apply(log10(x),1,mean) x2=x/xgeom x2.obj=log10(x2) CorCompare(cor(x2.obj[,1:10]),cor(log10(x[,1:10])), labels1=dimnames(x)[[2]],labels2=dimnames(x)[[2]], method1="centred logratio transformation",method2="log-transformation", ndigits=2) dev.off()