Given the $SVD$ of a matrix comprised of centered data points in $R$3, how do I find the line of best fit through the origin?

43 Views Asked by At

I'm slightly confused by this question - from my understanding the vectors u in $U$ are first principal components. In that case, would they already compose the line of best fit through the origin (since the data points are centered) with direction vectors in $V$^$T$? How would then I go about finding the projection of each point on the best fit line using the $SVD$?