Scholarlink

In e-commerce, accurately recommending product combinations that align with customer needs to be based on transaction history is crucial for increasing revenue. This problem, also known as “potential basket analysis,” requires identifying frequently purchased product sets. To address this, association rules are employed to calculate the probability of product group occurrences and predict customers’ subsequent choices. We propose new algorithms TUP FP-Growth and TUP-2 FP-Growth—improvements over FP-Growth, designed for rapid data processing without rescanning old data as the database grows. The TUP FP-Growth algorithm is based on the combinations of pairs of products. Initially, the algorithm will check and filter out products that satisfy the minimum support value. Subsequently, it examines the combinations of product pairs and yields results based on these two-item combinations. This new algorithm has been implemented alongside the original (FP-Growth and the primitive association rule mining algorithm Apriori) to compare their speeds on three databases of varying sizes. The results demonstrate that the improved algorithms perform faster, proving their efficiency and practical applicability.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

The Proposed Improvement to Frequent Pattern Growth Algorithm to Update Product Coordination Effectively in a Changing Data Environment

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Quoc Hung Nguyen,
  • Viet Phuong Truong

摘要

In e-commerce, accurately recommending product combinations that align with customer needs to be based on transaction history is crucial for increasing revenue. This problem, also known as “potential basket analysis,” requires identifying frequently purchased product sets. To address this, association rules are employed to calculate the probability of product group occurrences and predict customers’ subsequent choices. We propose new algorithms TUP FP-Growth and TUP-2 FP-Growth—improvements over FP-Growth, designed for rapid data processing without rescanning old data as the database grows. The TUP FP-Growth algorithm is based on the combinations of pairs of products. Initially, the algorithm will check and filter out products that satisfy the minimum support value. Subsequently, it examines the combinations of product pairs and yields results based on these two-item combinations. This new algorithm has been implemented alongside the original (FP-Growth and the primitive association rule mining algorithm Apriori) to compare their speeds on three databases of varying sizes. The results demonstrate that the improved algorithms perform faster, proving their efficiency and practical applicability.