Which data mining technique identifies associations such as customers with two or more vehicles likely to buy umbrella policies?

• A. Classification
• B. Association rule learning
• C. Regression analysis
• D. Cluster analysis

Answer: (B)

Association rule learning finds patterns that describe how items or events tend to occur together in data. In insurance data, it can reveal that customers with two or more vehicles are likely to buy umbrella policies, expressed as an if-then rule like “if two or more vehicles, then umbrella policy.” This captures a relationship between products, useful for cross-selling and understanding how policies relate to one another. The strength of such rules is measured by metrics like support (how often the rule appears) and confidence (how often the rule holds when the condition is met).

Classification, in contrast, would assign each customer to a category such as “will buy umbrella policy” or “will not buy,” focusing on predicting a single label for each individual rather than a relationship between items. Regression would estimate a numeric value, like predicted premium or claim amount, not an association between products. Cluster analysis groups customers into similar segments without explicitly identifying how different products relate to each other.