Association Rule Mining with a Memory-Efficient Apriori Algorithm: Optimized Transaction Pruning

The original Apriori Algorithm is known for its highly iterative process of constructing recurring patterns among mixed datasets for rendering Association Rule Mining; this led to the development of the Optimized Transaction Pruning (OTP) Apriori Algorithm to augment the memory utilization of the original Apriori Algorithm. While the earlier studies addressed computational memory cost concerns, handling extensive datasets effectively still requires higher enhancement. This paper introduces a novel procedure for Association Rule Mining (ARM) that addresses the issue of high computational cost in the Apriori algorithm by employing memoization and hash tables for transaction pruning. The proposed method, Optimized Transaction Pruning (OTP), significantly decreases memory usage while maintaining high accuracy through the implementation of the Memory Efficiency Metric for Association Rule Mining; the OTP architecture substantially reduces memory utilization, leading to a remarkable 58% reduction distinguished from the original Apriori Algorithm and a commendable 51% improvement over traditional transaction pruning techniques. Further, the validation of the effectiveness of OTP was tested across five diverse datasets while signifying its performance accuracy, which demonstrates an impressive 92% accuracy rate by utilizing F1 measures. The performance of OTP demonstrates the generalizability and robustness of the optimized algorithm, thus highlighting the viability of OTP in improving both memory efficiency and precision in ARM procedures, delivering a flexible solution for mining association rules in resource-constraint requirements.