Constrained smoothing splines are discussed under order restrictions on the
shape of the function m. We consider shape constraints of the type m((r))
greater than or equal to 0, i.e. positivity, monotonicity, convexity,.... (
Here for an integer r greater than or equal to 0, m((r)) denotes the rth de
rivative of m.) The paper contains three results: (1) constrained smoothing
splines achieve optimal rates in shape restricted Sobolev classes; (2) the
y are equivalent to two step procedures of the following type: (a) in a fir
st step the unconstrained smoothing spline is calculated; (b) in a second s
tep the unconstrained smoothing spline is "projected" onto the constrained
set, The projection is calculated with respect to a Sobolev-type norm; this
result can be used for two purposes, it may motivate new algorithmic appro
aches and it helps to understand the form of the estimator and its asymptot
ic properties; (3) the infinite number of constraints can be replaced by a
finite number with only a small loss of accuracy, this is discussed for est
imation of a convex function.