00380 6b4fc5cd24cdfcb1d8dc078dc50655

384

Obencbain

C    B

Figurę 10. Which Process Is Most Capable?

Table 3. Key Yalues from the CC Curyes of Figurę 10

Cumulative Capability	1 = 0	1=1	1 = 2	1 = 3
Process A	0.000	0.650	0.870	0.920
Process B	0.000	0.630	0.865	0.930
Process C	0.000	0.500	1.000	1.000

What Happens When Regret Is Not 2-to-l? Suppose now that your regret does not have a uniąue target value or is not continuous or does not increase monotonically with deviation from target. Especially when X can assume values on both sides of the target, there may be anywhere from one or two to an infinity of different X values that yield

the same regret, R(Xi) = R(X2),

the same regret index vaJue, I(Xi) = I(X2), and thus

the same cumulative capability, CC(I).

Poissonization of Regret

It can be extremely wastefiil of the information content of your measurements to simply convert them into pass/fail data via goal-posts regret. In this section, we consider an approach that enables us to use an