COMPARISON OF MULTIPLE REGRESSIONS IN MACHINING EXPERIMENTS

In machining research it is desirable that several variables should be investigated simultaneously, but in such a way that the effect of each one can be reliably evaluated. This paper describes how the significance of the correlations between the individual variables and the dependent variable may be tested and also how several multiple regressions may be compared with respect to variability, parallelism and displacement in a composite analysis. The technique is demonstrated on results of turning experiments in which the cratering behaviour of two carbide compositions is compared. The results of the regression analysis are discussed critically.