Apriori algorithm is one of the more practical and commonly used algorithms today. It’s a sequence used for dataset mining and is truly helpful in forming associations within the dataset. The fact that it can be used to make assumptions about the relationships within the group makes it highly significant in identifying trends.
This algorithm was developed in 1994 by Srikant and Agarwal. The idea is to come up with a sequence that would determine the association of an item to another through a series of transactions. This is opposed to what the then-current market has to offer. There are other algorithms in place in doing this like Winepi or other sequencing with no timestamps but there are none that would incorporate the use of transactions. In doing this, the whole database is composed of transactions wherein each transaction, comprising a set of items, becomes a subset of the whole database.
The apriori algorithm works this way. A subset is evaluated in comparison to another subset. They do this by extending a subset and branching it out on a per item basis. Once an item extension is deemed successful, it will continue to extend its scope and test another item in the subset. It only terminates when the extension fails. This applies to all the subsets, meaning every item will be pitted against each other. This is useful in determining how close the relationships are with one another.
A more practical application of apriori algorithm is the database used in supermarkets or department stores. Every purchase becomes a subset and the algorithm works to determine the consumer behavior. It’s highly essential in doing the market basket analysis which is a theory that if one purchases this item, it is highly probable that one will purchase a set of other items as well. The algorithm helps determine the frequency of purchases and the trends involved.