Helper to compute confidence ellipsoids
Computes the quantile of the Mahalanobis distance needed to draw confidence ellipsoids.
ConfRadius(model,prob=1-alpha,alpha)
model |
A multivariate linear model |
prob |
The confidence probability |
alpha |
The alpha error allowed, i.e. the complement of the confidence probability |
Calculates the radius to be used in confidence ellipses for the parameters based on the Hottelings T^2 distribution.
a scalar
K.Gerald v.d. Boogaart http://www.stat.boogaart.de
data(SimulatedAmounts)
model <- lm(ilr(sa.groups)~sa.groups.area)
cf = coef(model)
plot(ilrInv(cf, x=sa.groups))
for(i in 1:nrow(cf)){
vr = vcovAcomp(model)[,,i,i]
vr = ilrvar2clr(vr)
ellipses(ilrInv(cf[i,]), vr, r=ConfRadius(model, alpha=0.05) )
}Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.