# What is an “If-Then” rule?

Can someone give a concise, layman's explanation of an "If-Then" rule (as in rule-based systems). I am finding this term used frequently without anyone really defining it properly.

-
Do this question and the answer refer solely to machine-learning? –  Henrik Sep 17 '10 at 8:12
@Henrik Yes, as a base of rule-based classifiers. –  mbq Oct 12 '10 at 13:41

It is just a simple classifier; so simple that it is better to explain it by example. Let's say you have a 3 class classification problem and information system with 4 continous predictors $X1,\ldots, X4$. Now you can define simple rules like:

IF X1<3.45 THEN Class1
IF X3>7.2 THEN Class2
IF X2<2.11 THEN Class1
IF X2<1.2 THEN Class3
...


and this is it; the final decision is produced by voting. Structures like this can be trained by various methods, probably the best known are rough sets.

-
So a set of "If-Then" rules is basically a forest of single-level decision trees (i.e., single binary splits on a single predictor)? –  user1121 Sep 16 '10 at 19:30
@user1121 Yes, but only in case of binary classification. –  mbq Sep 16 '10 at 19:35