Association

As consumers, most of us have encountered recommendation engines where a movie is recommended based upon prior viewing habits or books are recommended based upon prior purchases (or even based simply on prior viewing behavior). Association Rules analysis is often described primarily as kind of market basket analysis, but while it is strongly associated with retail data it can be applied in other areas. For example, in the area of predictive maintenance, a pair of part failures might be frequently associated with a third part failure even though the third part has not yet shown evidence of trouble. Modeler supports three different Association nodes: Apriori, Carma, and Association Rules.

The fourth algorithm, Sequence, is a bit different. It takes either the time-date stamp, or a simple ranking in time, into account. The transactions in Association Rules often occur at the same time. For instance, on a receipt at a grocery store many items might be listed. Hot dogs and hot dog...