Association rule learning algorithm
Association rule mining is more useful for categorical non-numeric data. Association rule mining is primarily focused on finding frequent co-occurring associations among a collection of items. It is sometimes also called market-basket analysis.
In a shopper's basket, the goal is to determine what items occur together frequently. This shows co-relations that are very hard to find from a random sampling method. The classic example of this is the famous Beer and Diapers association, which is often mentioned in data mining books. The scenario is this: men who go to the store to buy diapers will also tend to buy beer. This scenario is very hard to intuit or determine through random sampling.
Another example was discovered by Walmart in 2004, when a series of hurricanes crossed Florida. Walmart wanted to know what shoppers usually buy before a hurricane strikes. They found one particular item that increased in sales by a factor of seven over normal shopping days; that item was not bottled water, batteries, beer, flashlights, generators, or any of the usual things that we might imagine. The item was strawberry pop tarts! It is possible to conceive a multitude of reasons as to why this was the most desired product prior to the arrival of a hurricane–pop tarts do not require refrigeration, they do not need to be cooked, they come in individually wrapped portions, they have a long shelf life, they are a snack food, they are a breakfast food, kids love them, we love them, the list goes on. Despite these obvious reasons, it was still a huge surprise!
When mining for associations, the following could be useful:
- Search for rare and unusual co-occurring associations of non-numeric items.
- If the data is time-based data, consider the effects of introducing a time lag in data mining experiments to see whether the strength of the correlation reaches its peak at a later time.
Market-basket analysis can be applied to the following areas:
- Retail management
- Store management
- Inventory management
- NASA and environmental studies
- Medical diagnoses