Using regression techniques in cost analysis Part II †

Publisher Website
Google Scholar
Abstract
Regression techniques provide the cost analyst with a valuable analytical tool. This paper describes means of expanding the utility of regression analysis applications in this area. Part I of the paper presented basic definitions and developed measures to assess the accuracy and sensitivity of a regression formula. In particular, it was shown that the Association index, rather than the more commonly used R1, is the pertinent measure of formula sensitivity. We also described several other aspects of evaluating formula reliability—the confidence level, variability in the data, parameter stability, autocorrelation—and discussed indexes especially developed to quantify these factors. Part II deals with the application of regression techniques to such areas as shop machining costs and long-term overhead relationships. The selection of variables and observations for a model, and the use of non-linear regressions, are discussed. Several examples from aerospace, ship construction, and other manufacturing industries illustrate the discussion of complex non-linear formulae which includes several situations characterized by cost-improvement curves. The overall reasonableness of a formula is then discussed in terms of data comparability, statistical adequacy, operational relevance, and prediction potential, and the paper concludes with further comments on intercorrelation of variables.

This publication has 0 references indexed in Scilit: