CLTD-LP: an optimized top-down clustering approach with linear prefix trees for scalable frequent pattern discovery in large datasets

The extraction of frequent itemsets and association rules is a fundamental challenge in data mining and holds significant importance within the field. Mining techniques utilising Linear Prefix (LP) growth association rules employ a bottom-up methodology that necessitates a conditional pattern base and a conditional LP-tree for the extraction of frequent itemsets. This research proposes a Linear Prefix tree utilising a Top-Down Approach with Clustering (CLTDLP) method, to address the limitations of the current LP-growth algorithm. The suggested CLTD-LP algorithm employs a top-down methodology; it generates a subheader table to effectively mine common items, rendering the CLTD-LP algorithm more advantageous as it does not create a conditional pattern base or LP-tree. The proposed methodology enhances the algorithm’s efficiency regarding execution time and memory utilisation. Overall, across three benchmark datasets, the proposed CLTD-LP algorithm consistently achieves better average reduction in terms of runtime and memory than the existing LP-growth, Ordered Frequent Itemsets Matrix (OFIM) and Simple and Scalable Frequent Itemset Mining(SSFIM) algorithms.