On the determination of the parameters in an empirical formula
William Edwards Deming · Proceedings of the Physical Society · 1930
The writer reviews the method of least squares and gives reasons for recalling some cautions regarding its use that were mentioned by Stewart and Uhler several years ago. Examples are worked out to illustrate the discussion. The method recently mentioned by Awbery is compared with those of least squares and zero sum, from the practical point of view. Judging from the few examples given, the method of zero sum appears to be the quickest and its results to be well in accord with common-sense judgment. The amount of computation involved in the application of the various methods must be viewed in the light of the short cuts and elaborate calculating machines now available, since these may lighten the labour for one method more than for another. Attention is called to the fact that the method recently expounded by Awbery is the one given by Cauchy in 1847.