No agent wanting to close a sale would utter such a seemingly conflict
ing statement. One of the problems with [this method] is that the trans
formation from [(p and q) then r] to [(p then r) or (q then r)] shifts our
focus from one rule to what appears to be the union of two (or more)
rules. In addition, since each of these alternative rules can contain dif
ferent consequent subsets, they seem to be contradicting each other:
either my premium is high OR my premium is low. How can it be both?
For many of you, the counterintuitiveness of this method may be a stum
bling block but if so, persevere — it’s definitely worth the effort if you find
yourself working with large rule bases.
Fuzzy Inference and the Combs Method
When using the Combs method, rules are processed as normal. To clarify,
let’s work through an example with the same figures used in the example
shown earlier in the chapter: 200 pixels for distance and 8 for ammo status.
Calculating the result of each rule we get:
Target_Close ® Undesirable (0.0)
Target_Medium ® VeryDesirable (0.67)
Target_Far ® Undesirable (0.33)
Ammo_Low ® Undesirable (0.22)
Ammo_Okay ® Desirable (0.78)
Ammo_Loads ® VeryDesirable (0.0)
454 | Chapter 10
The Combs Method