The support, confidence, and lift are the important parameters to be set that define the output. Based on the business case, we set them accordingly. Let's see them in detail.
Support is an important measure in the process of extracting the association rules. It defines how often a rule is applicable in the dataset. For example, let's consider rule number 23 in the previous session, where we have {sex: "Male," native: "United States"}
in lhs
and {capital-loss: None}
in rhs; here, it has support
of "0.5661
," which means that the number of transactions containing {sex: "Male", native: "United States", capital-loss: None}
is about 56.61% of the total number of transactions.
Generally, the rules with very low support are neglected because they would have occurred mostly by chance, are not significant, and of no interest to the business as the chances of occurrence are very rare and not worth monitoring. However, in rare scenarios, when the number of transactions...