A Systematic Review of Frequent Itemset Mining Algorithms and Their Capabilities

EasyChair Preprint 12251

Abstract

FIM (frequent itemset mining) is a key data analysis task. that is important for large businesses decision-making. The efficient mining of frequent patterns from transactional databases has been proposed using a variety of FIM methods. This work gives a thorough evaluation of numerous conventional and parallel FIM methods, outlining the benefits and drawbacks of each.The classic Apriori algorithm creates candidate itemsets using breadth-wise searching. Due to the numerous database searches, it has a significant memory and calculation time consumption .On the other hand, FP-Growth compresses the whole database into an FP-Tree after just one scan, but creating the FP-Tree can take some time for large databases.Modern algorithms like MMFI and Max-IFP use FP-Arrays and FP-Matrix to get over these restrictions, which leads to more effective memory usage and quicker frequent pattern generation. When compared to FP-Growth, the COFI algorithm uses less memory since it takes a pruning strategy. Enhanced variations, including Eclat and CP-Miner, introduced strategies to address memory consumption and runtime efficiency. Additionally, emerging paradigms like parallel and distributed FIM algorithms have gained prominence with the advent of big data. However, challenges remain in terms of scalability, noise handling, and pattern diversity. This comparison study makes it clear that a more potent and scalable FIM algorithm is therefore essential. This systematic literature review aims to provide a comprehensive analysis of various FIM algorithms, highlighting their strengths, limitations, and applications across different domains. Future research will concentrate on improving the suggested method and running comprehensive trials to assess its effectiveness and scalability.

Keyphrases: Apriori, Association Rule Mining, Eclat, FPMAX, Frequent Pattern Mining

@booklet{EasyChair:12251,
  author    = {Anshu Singla and Parul Gandhi},
  title     = {A Systematic Review of Frequent Itemset Mining Algorithms and Their  Capabilities},
  howpublished = {EasyChair Preprint 12251},
  year      = {EasyChair, 2024}}