What data type must a data set's attributes be in order to implement association rules in RapidMiner?

• A. Binominal
• B. Numeric
• C. Text
• D. Categorical

Answer: (A)

Association rules look for how items co-occur in transactions, which is most naturally represented with binary indicators: each attribute corresponds to a specific item and its value simply shows whether that item is present or not in a transaction. In RapidMiner, this means the data should be binominal—two-valued attributes (yes/no, true/false). With this binary encoding, the algorithm can easily count how often items appear together across transactions to compute supports and confidences for rule mining.

If you have numeric, text, or multi-valued categorical attributes, you’d first convert them into binominal (binary) attributes or a one-hot encoded form so each potential item is represented as present or absent. Numeric attributes aren’t directly interpreted as presence of an item, text needs tokenization or feature extraction, and multi-valued categories don’t fit the simple present/absent framework without such preprocessing.