This self-study problem illustrates how nonlinear regression can be used to estimate the learning curve rate,
given information on output levels and average processing times. To illustrate, we reproduce below the cumulative
output ( X ) and the average time per unit ( Y ) for the SofTech Inc. illustration in Exhibit 8A.2 (page 291).
Using regression analysis, we now show that this learning curve data is consistent with the general learning
model. The general form of the learning model is Y aX b where a the time required for the first unit and
b the learning index. Using a log transform for Y, the general learning model can be shown in the equivalent
log-linear form (see footnote 13 in Appendix A): log( Y ) log( a ) b log( X ). The log transform is
available in Excel in “Insert Function” under the Formulas tab.