 |
Results of classification algorithms can also be rules like for Apriori. But for classification rules the right hand side has only the class attribute, while for association rules there is no such restriction, i.e. right hand side of the rules can have any number of attributes, including attributes other than the class attribute. |
|
|
Association rules can find a rule like X Þ c, where c is one of the classes. This rule is very much like a classification rule, but this rule gives the information about what (X) best describes class c. While the classification rule, X Þ c differentiates between class c and the other classes. |
|
|
Clustering algorithms do not use information about the class labels, while association rules uses them (it just treats the class attribute as the other attributes). |
|
|
For X Þ Y, if a cluster is defined by using only Y, different clusters can be formed sharing the same description, X. |