On an alternative solution to the vector spline problem
Yee and Wild have presented additive extensions to multivariate regression models including generalized estimating equations. The back-bone of their methods was a vector spline, which was solved by direct solution. A back-fitting solution to the vector spline problem is presented here, with the implication that existing generalized additive model software can, with moderate effort, be modified to implement multivariate models.
No Supplementary Data
No Article Media