Let \(y = [ y_a^\top \ y_b^\top ]^\top\) be a random vector with finite second moments and let \[
\tilde{y} = f +
\begin{pmatrix}
G & 0 \\ 0 & H \end{pmatrix} y,
\]
where \(G\) and \(H\) are non-singular. Show that the canonical correlations of \(y\) are equal to those of \(\tilde{y}\).
The dataset nhanes.RData
contains demographic, diet and health information on a sample of the US population. Analyze the relationship between the 10 dietary variables “DXXXX” and the health variables “BXXXX”.
There are a variety of ways to address Question 2, so I expect a lot of heterogeneity in the data analyses (but hopefully not in the overall conclusions).