Aw. Thomas et al., CONTRASTING SLIDING POLYNOMIAL SMOOTHING AND POLYNOMIAL REGRESSION INDATA-ANALYSIS, Transactions of the ASAE, 38(6), 1995, pp. 1935-1938
Data processing is a necessary step in organizing and summarizing nume
rical results of experimentation. Various methods have been proposed a
nd utilized. Results of agricultural research as well as other natural
resource research may depend upon the methods of data analysis and mo
dels used as well as upon the experimental data themselves. Two types
of data analysis have been contrasted in this article: sliding polynom
ial smoothing and polynomial regression. Smoothing is shown to be supe
rior in adherence to the data, in subduing polynomial bounce, and in m
aintenance of required symmetry. It is shown that adequacy of data ana
lysis cannot always be measured by goodness-of-fit.