A Comprehensive Review on Frequent Pattern Mining Algorithms in Data Mining

Frequent Pattern Mining (FPM) is a basic data mining method to find recurring patterns in data that is used in a large number of applications in areas of market basket analysis, recommender systems, and fraud detection. Three basic types of FPM algorithms, i.e., Join-based, Tree-based, and Vertical-based algorithms, are discussed in this paper. Join-based algorithms employ iterative candidate generation, Tree-based algorithms employ compact tree representations to support efficient generation of patterns, and Vertical-based algorithms employ a vertical layout of the data to obtain better results using rapid intersection operations. By going through classic algorithms such as Apriori, FP-Growth, and Eclat, this paper introduces overviews of mechanisms, advantages, and limitations of each type to present a complete view of existing methods in frequent pattern mining.