It turns out that very simple association rules, involving just one
attribute in the condition part, often work disgustingly well in
practice with real-world data. Suppose in the weather data, you wish
to be able to predict the value of `play`

. The idea of the OneR
(one-attribute-rule) algorithm is to find the one attribute to use
that makes fewest prediction errors. For example, consider
`outlook`

:

if outlook = sunny then play = no .. makes 2 errors in 5 records if outlook = overcast then play = yes .. makes 0 errors in 4 records if outlook = rainy then play = yes .. makes 2 errors in 5 recordsfor a total of 4 errors in 14 cases. Likewise,

if humidity = high then play = no .. makes 3 errors in 7 records if humidity = normal then play = yes .. makes 1 error in 7 recordsalso for a total of 4 errors in 14 cases. The other two attributes each produce 5 errors at best, so the OneR algorithm chooses at random betweeen using

`outlook`

and `humidity`

as the one decisive
attribute. The algorithm is:
For each attribute A: For each value V of that attribute, create a rule: 1. count how often each class appears 2. find the most frequent class, c 3. make a rule "if A=V then C=c" Calculate the error rate of this rule Pick the attribute whose rules produce the lowest error rate

This version: 2000-10-30