Model robustness of Hoel-Levine optimal designs (Q1067721): Difference between revisions

This paper considers the performance of the least squares estimator when it is combined with the \textit{P. G. Hoel} and \textit{A. Levine} optimal design [Ann. Math. Stat. 35, 1553-1560 (1964; Zbl 0127.103)], when dealing with extrapolation of a function which can be approximated well by a polynomial. It is compared with an optimal spline extrapolator using mean squares error as the criterion.